Chinese AI founders are often different from American ones. They tend to stay away from the media and put all their energy into building models. They seldom get the chance to fully share their vision for the industry. So this talk, which lasted nearly four hours, gives us a rare look into his thinking.

I don’t usually like ideas that are too abstract, but Liang’s understanding of AI is almost like Daoist philosophy. He does not see AI simply as a tool to build a monopoly. Instead, he sees it as a huge wave that is changing human history, a tide that no single company can ever own. That is why he believes the best attitude is to hold back and be kind: you should set limits on your own gain, earn only a fair profit, and share openly to reduce pushback. In his heart, the real goal is to reach AGI. All other business wins are just side products, natural results as the technology moves forward, not the main goal. One interesting point: he thinks the biggest key to reaching AGI is not finding the very best people, but keeping the team stable.

Liang believes that the lead among top American models, such as OpenAI, Anthropic, and Google, is cyclical and will not last. He says Anthropic’s early edge in coding agents will soon fade. When talking about the gap between China and the US, he says America’s lead comes only from having more computing power. He thinks talent is spread around the world by chance, and China has no real shortage of it. In his eyes, American models keep getting bigger because they have plenty of resources, and their basic business idea is to lock up the market with closed models and high profits. But this pursuit of very high profits is weak in terms of strategy, because it will surely be beaten by those who are happy to earn only a fair profit.

He believes China has built-in advantages in cost and user experience, since American companies have no strong reason to push costs to the very limit. He pictures a future where competition looks like that in other manufacturing fields: China will play the role of offering cheaper services in a systematic way. At the same time, he thinks Nvidia’s CUDA ecosystem wall is being broken down by new technology, and that building a new ecosystem for Chinese chips is a historic chance. Once the limit on how many chips can be made is overcome, the gap in basic computing power between China and the US will go away.

In his view, the future AI world will never be a pyramid controlled by one or a few giants. It will be a system based on holding back, one where everyone gains and lives together. First, the number of companies working on base models will shrink; in the end, only three or four need to compete fully, and any that try to make very high profits will be pushed out. Second, he insists that open source is the base of the ecosystem. He believes a pricing plan that earns only a small profit, such as getting back the hardware costs in ten months, can let open source and business work together without conflict. This will push more partners to build apps on top of open source models. Finally, DeepSeek itself will focus strongly on the main road to AGI: mainly language models, chains of thought, agents, and continual learning. It will actively leave aside areas like video generation and world models, giving those business opportunities to the whole ecosystem and to society. By making only the most important tech advances, the company hopes to help the whole ecosystem grow, with an attitude of not fighting or competing.

For other highlights, I would recommend ’s post The DeepSeek Doctrine

Below is the transcript I made with the help of AI translation:

Liang Wenfeng Investor Exchange Meeting · Audio Transcript

Audio “deepseek 0520.m4a”, total length approximately 3 hours 44 minutes. This transcript was auto-transcribed from speech recognition and organized by AI, without distinguishing speakers; the content in brackets indicates the audio timestamp position; individual proper nouns and numbers may contain recognition errors, please refer to the original recording as authoritative.

[00:00:01]

...as well as our company’s other colleagues, when we first started this company, our original intention was not that I ultimately wanted to make a certain amount of money, to go to the capital markets, to IPO, or anything like that. So we didn’t have that original intention.

The first few dozen people didn’t think this way at all. If someone thought that way, they wouldn’t have come. So overall, we’re doing this with tremendous goodwill toward the world, and then we feel this is useful for humanity—this is something beyond money.

Of course, later on, after discovering that the benefits of this thing are enormous, there were other temptations—that’s a separate matter. But our original intention when we set out, our vision, and the vision we’ve maintained until now, was not done in a way that maximizes commercial interest. I think this point is quite crucial.

About twenty years ago, the person I admired most in management was Jack Welch, the former CEO of GE. Looking back now, most of what he said may already be wrong, but he got one most important point right: the most important thing about a company is its vision.

Managing a large company relies not on your rules and regulations, but on vision. What is vision? Vision is not a slogan hanging on the wall—vision is how you do things, not how you talk, that is, how you actually operate. Anyway, I forget Jack Welch’s exact words, but that’s roughly the meaning.

So how do we manage so many people, how are we organized? Actually, we have no organization—it’s vision-driven, organized by a vision. We have no organization.

This has advantages and disadvantages. In the future we’ll figure out how to build on strengths and avoid weaknesses, but this is our characteristic. We don’t operate in a way of “I need to achieve some KPI, no assessments”—only vision.

This vision isn’t even written down, it’s not written out, nothing has ever been written down. This vision exists in our methods of doing things, in our attitude toward the world. Perhaps everyone in our company understands this vision differently, perhaps each person’s vision is somewhat different, but on a broad direction we are aligned.

I think it’s still about having tremendous goodwill toward the world, and then wanting to accomplish something. This is what we use to organize ourselves.

Next I’ll speak first, and after I finish everyone can ask questions. I’ll probably talk about the subsequent matters revolving around this vision. This vision is real, not made up—we truly think this way, truly do it this way. Otherwise you can’t explain many of our actions.

Why do we insist so much on open source because of this vision? Because this vision itself requires open source. Without this vision, you can’t organize people.

For example, Zhipu also open-sources, but Zhipu’s open source is different from ours. Zhipu’s open source has a forced feeling—they feel it’s not their true intention, but for us, this is our true intention.

Then on the matter of open source, we thought it through very clearly from the start. First, the first reason is vision; second, we believe that to succeed commercially with AI, open source has benefits.

This sounds a bit contradictory, a bit counterintuitive, because historically, open source and commercialization have been in conflict. But I think AI is different from before. Because historically, a software company’s market might be just a few billion dollars a year, and once open-sourced, it’s gone—maybe only tens of millions or a few hundred million left.

But AI is big enough—ultimately it might occupy, say, ten percent of human society’s GDP. That’s actually a very large number. One person monopolizing this thing—you can’t monopolize this thing, you must share with others, otherwise you definitely can’t survive.

This is different from previously open-sourcing a piece of software, because that software’s market wasn’t that big. But AI is simply too big. If we want to monopolize this benefit, then we’re bound to be abandoned by history. I think the main point is that this is an objective law, this is a view of history.

It’s not that if I don’t open-source, I can monopolize this market—that theoretically doesn’t conform to objective reality. You’ll definitely encounter many obstacles, there will definitely be other methods to stop you from achieving this goal.

In this situation, I think you don’t necessarily have to follow traditional business thinking. You need a mechanism to ensure that the benefits you yourself can obtain are limited—only then do you have a chance of succeeding. Restraint is needed, I think restraint is needed.

If we want to succeed at AI in our hands, first of all, I think restraint is needed. You can’t think that some percent of humanity’s GDP all belongs to me, or that some percent of China’s GDP all belongs to me. The more you think this way, the less you’ll succeed.

So from the beginning we felt restraint is needed. The more restrained you are, the more likely you are to succeed at this thing. This is a commercial consideration, though it’s a macro-level consideration.

I think this conforms to intuition, at least it conforms to my intuition, or at least I truly think this way. We don’t have many other advantages, we have no special skills, we’re not richer than others, nor are our personnel better than other companies—we really aren’t.

Think about it, when we founded this company two years ago, we didn’t have much money, we didn’t have many cards [GPUs], we had no fame, no influence—we were just a group of very ordinary people.

[00:11:49]

We really are just a group of ordinary people. If there’s a narrative I like, it’s a group of ordinary people accomplishing extraordinary things, rather than a group of geniuses accomplishing extraordinary things. This is closely related to our restraint—it’s of one piece with our restraint and our vision.

So, will open source and commercialization conflict? I think on this matter of AI, if you’re not restrained, you won’t rise. Open source is part of restraint, and our restraint is manifested not only in open source but in many other aspects too. But overall, we don’t need to worry about open source, don’t need to worry about restraint.

The more restrained you are, the easier it may be to succeed, or at least so far it’s been borne out, so far it can be explained. Otherwise there’s no way to explain why we could succeed: we had no weapons, our starting point was very low, our resources were very few, and our people are actually just a random group of ordinary people. I myself am just a university graduate, and not from the most top-tier school either.

This restraint is also part of our vision. AI is too big, the benefits too large. We are very restrained—as long as we can succeed, ultimately the benefits will be enormous. However you divide up even a little, the benefits are enormous, so right now there’s simply no need to consider which portion of these benefits to take, or how to take it—I think there’s simply no need to consider this, because the benefit is already big enough.

You only need to take a tiny bit and it’s already more than enough. So as we said before, we only take a reasonable profit—it depends on your willingness, not the size of profit—this is different. This isn’t our API pricing. Our API pricing considers a reasonable profit—roughly, we go to the market and buy a batch of equipment, and recover the cost in ten months. I think this is a reasonable profit.

Under current circumstances, considering you have risk, plus upfront investment and so on—if a server, we financially amortize over three or five years, but commercially, we feel roughly ten months to recover cost, we feel that’s enough, OK, we feel it’s enough. So this is the logic of our current API pricing. Our V3.2 Flash and the others all recover equipment cost in ten months.

This is our standard. It’s actually not profit-maximizing—if it were profit-maximizing, we should set prices higher. Because in this price range, user demand is inelastic, meaning if I cut the price in half, or if I raise the price by double, the token consumption doesn’t differ much.

If I make the price twice as expensive, my total revenue would be close to double. Wait a moment, let me check. Ah, great.

Let me tell you a story—it’s about our DDCP, one of our models. At first we worried demand would be too high, so we set the price relatively high at first, and people on the team weren’t very happy. Later I brought the price back down, down to a quarter, and everyone was happy.

I think this is our true thinking. It’s the vision I mentioned earlier—we still want this thing to be useful to people, rather than us making the most money—rather it’s that under the condition of being able to make a reasonable profit, everyone can afford it. I think our company’s other people’s thinking too—when we cut prices, many people in the company group cheered, everyone felt very happy.

Because this is the purpose for which we spent so much effort, so much care making this model well. The purpose is to be very cheap, very effective, letting everyone fully use it. We just feel this is happy, this is our motivation, this is our vision, this is the consensus that allows our company to come together to do this thing—this is our company’s internal consensus.

This point should be relatively special, because price cuts like this are certainly not a good thing for our competitors—they definitely don’t cheer. Because your revenue, your ARR—if you cut in half, ARR drops by half. Right, this is a place where we’re different.

We feel this is enough. From our company’s internal perspective, I recover cost in ten months—commercially I’m already very satisfied. For the company’s external perspective, we also feel this price is one everyone is happier and more willing to see—it’s win-win for everyone, win-win for the company, society, and everyone.

I think, OK, someone just left a message on screen saying ten-month payback is too high a profit. Indeed there’s still room for price cuts, indeed still room. There’s also room for optimization in the models, so overall there’s quite a lot of room for price cuts.

But this cost, ten-month payback, we ourselves can achieve—other companies can’t. Like maybe Alibaba or Tencent—they don’t have our optimization, their cost should be several times higher than this. There’s still a lot of optimization work here.

I just said why we don’t keep cutting prices—it’s because it’s inelastic. That is, if I cut prices further, demand won’t increase more, or if I cut prices further, demand increases very little. Because at this price everyone can already afford it, everyone feels this price is satisfactory, and won’t stop using it because the price is expensive.

So cutting prices—first, the company won’t have more revenue; for society there’s no more value either, because everyone is already satisfied with this price. Even if you lower the price further, it won’t increase society’s sense of happiness much. Right, OK, but on this issue just now, on this matter of pricing, we’re definitely not...

[00:25:32]

...taking company revenue highest or profit highest as our starting point. This is part of our restraint, because short-term, if your price is a bit higher, you might have a bit more revenue; but long-term, it’s really hard to say. Because I think restraint is a strategy.

For me, restraint is a strategy. It lies in the fact that sometimes you can give up some things to exchange for more of other things. The matter of not open-sourcing is actually the same—it can be seen as our pressure, or as our concession of benefits.

First, this concession—for our company internally, we’re very happy, everyone is happy, employees feel a sense of accomplishment, and we’ll have cohesion because of it. And this concession is good for society, society is also happy, other peers or ordinary people will be happy. So this restraint, my understanding is, this restraint from a long-term perspective can increase our probability of achieving AGI.

When considering a matter, I have no doubt that AGI will have enormous commercial value. So on this basis, my priority consideration is not how I add a bit more share, how I take a bit more share—my priority consideration is how I increase my probability of succeeding.

This restraint is perhaps reflected in many other aspects too. For example, last Spring Festival we suddenly had many users, but we didn’t pursue retaining these users, or monetizing these users, or grabbing these commercial benefits, cashing in on the users.

We didn’t grab users, didn’t make money, but we worked very hard to figure out how to serve users well. We wouldn’t have the thought of “I want to become the next super app, and then I want to compete with someone, I want to become the next ByteDance, become the next Tencent”—we have no such thoughts at all.

We could do it this way, but we didn’t. My understanding is, this is also part of restraint. You shouldn’t think you have to profit from everything—it seems like once you have users, it seems you could become the next ByteDance, and then you just gobble that up.

I think this is workable commercially, it’s possible. If last year we had used a large sum of money to grab users from ByteDance, that would also be a way to play. But we chose a very restrained approach—that is, I won’t compete with you over this thing, because there are still watermelons behind, and what’s in front is maybe just sesame seeds.

I shouldn’t grab all the sesame seeds. Of course, maybe these sesame seeds are relatively large, but I think the AI behind might make what’s in front seem small. Looking at it now, our not pushing hard on the C-end last year was probably correct. Because you can see that behind there really are bigger watermelons, and what’s in front really is just some small sesame seeds.

If last year I had a lot of money and made this thing very big, what would be the benefit? You wouldn’t have gained anything. These are my true thoughts, because I think the AGI opportunity behind should be very large, the AGI opportunity behind is always very large.

I don’t even need to consider whether I’ll occupy a position in it at that time, or what my business model in it will be—we simply don’t need to consider it. As long as there’s such a big commercial opportunity, you’ll definitely have a way. So for the sesame seeds in front, we’ll also pick some up, but we just pick up a bit casually—we won’t stop, or treat it as an important matter.

So last year’s C-end DAU and such—I think it’s probably a small matter. But we picked it up too, we maintained user usage at a relatively low cost, because it might be useful later. Although right now we don’t know what use these users have, right now it’s a pure cost expenditure, but it might be useful later.

Since it can be obtained casually, we’ll pick it up along the way. Including this year, it’s very likely that our ARR revenue from API or AI also has an opportunity. That is, if this demand can continue to expand—if it keeps expanding, and we can buy more GPUs—then ARR reaching several hundred million dollars is very possible.

If AI can reach a billion dollars, then basically my company’s cash flow could turn positive, could cover my R&D expenses, could cover all my expenses. So this is also possible, but we haven’t made it a priority.

This we’ll do, but I think this is an important matter. It’s not our first priority consideration, or not what we truly care about today. The bigger opportunities should still be behind—the opportunities in front, including last year’s C-end and this year’s B-end—I think this needs to be done, needs to be done well, but this is not our goal.

Or rather, most people in our company don’t think this is a very important matter, don’t think this is a matter equally important compared to AGI.

The open source part I can say more about, because many previous questions were about open source. First, I think we will open-source, and then even our strongest model will probably be open-sourced too. Because I don’t see the benefits of closed source, I don’t see the inevitable benefits. ByteDance’s model is closed source—what benefit does it have? I don’t see any benefit.

Even if a model is open-sourced, even if you tell others everything, the threshold is still very high. For others to actually use it, the threshold is very high. For them to use it, it’s very hard; secondly, for them to use it and get the cost very low too, is also very very hard, not that easy.

It’s not that once I open-source, they can easily achieve the same deployment cost as me. There’s still a lot of work to be done here. Although the principles of this work are all understood, not every company is willing to, or has the willingness and ability to organize manpower to reach this goal.

I’m quite used to this point too. It may just not be good at doing this thing, because the resistance is too great. Controlling this cost is very hard—it has many management and physical constraints. This is also the advantage of startups, because if a startup is too small, you don’t have the strength to do this thing; if you’re a big company, you find it hard to organize.

[00:38:36]

This matter has difficulties everywhere, so this belongs to the sweet spot of a company of our scale, the sweet point. Then if we were bigger, we might have other problems; if smaller, our friends’ strength would be insufficient.

So, as for open source, I think we should set pricing. Currently we should recognize, we won’t force people. Because for the pricing model, I won’t charge a very high fee—I’ll probably charge based on ten-month payback. At ten-month payback, it’s already enough to keep competitors...

At my ten-month payback, I can make independent third-party deployment unprofitable. The third party can’t do it, can’t achieve this cost—they definitely can’t.

So open source won’t affect my revenue. Of course, if I wanted to earn a hundredfold profit, then open source is...

“I can hear you, but the video seems to have dropped, boss.”

“Just now maybe a phone call came in.”

So open source, I think has no impact whatsoever on our business model. The premise is that we only earn sixfold profit—ten months to recover cost roughly corresponds to sixfold profit. Under the condition that we only earn sixfold profit, open source won’t have any impact.

But if you wanted to earn a hundredfold profit, then open source would indeed affect your earning a hundredfold profit, because third parties would deploy—they might be at twentyfold cost, which would be lower than yours.

Is this model sustainable long-term? I think it is. Under our vision, I think open source is sustainable long-term, or we intend to do it this way. You can consider it as having restraint, which is also giving you longer-term benefit.

This strategy can give us more opportunities technically at the front, and our probability of achieving AGI is also greater. We’re more at ease.

Think about it, we don’t even need to work overtime, because it’s just not that hard. But for other companies, it might be very hard, because you’re thinking about too much.

It’s actually not hard, simply not hard. At first it’s a... From the outside it might look like we chose a very hard model—we do research, we do the hardest things, seemingly a hard mode. But actually we gave up a lot in other areas, which lets us still be very strong, still doing it very easily.

So on this matter of open source, actually my judgment is that it’s sustainable. There’s no conflict between open source and commercial paid services—the premise is that under sixfold profit there’s no conflict.

Sixfold profit looks high, but actually it’s not high. Because now with AI efficiency this high, right now a reasonable profit might just be this much. In the future it might drop to, say, fourfold, threefold—I think that’s already... any lower and it couldn’t drop further. But it will still have very large profit.

Just looking at selling API alone—but I don’t think selling API is that attractive. But on this matter, you can explain, no, I don’t see any conflict.

Then I’m also not worried about others deploying our models and then competing with us—not worried at all. We even hope they can deploy it successfully.

We provide as much help as possible to the open source community, assisting everyone to deploy our models. I’m not worried they’ll steal this business from me, because this market is big enough. I’m only worried they can’t deploy it—that they got some details wrong, made the effect relatively poor, or that their cost is relatively high.

Right, there’s no conflict here.

Then last year, when there was this To B business, a question asked more often was: our C-end—if I open-source, then won’t the C-end conflict with my C-end again? Because I don’t have a traffic advantage, or Tencent itself has lots of traffic—if they deploy our open source model, they’d take over all the C-end users, taking away my C-end users.

But actually it won’t, for many reasons.

Then the question is: is the open source model we give the same as the model we deploy ourselves? It’s the same. We won’t open-source a slightly worse model and then use a better model when we deploy ourselves—we won’t do that, it’s the same.

This also shows there’s actually no conflict. All of last year, on the C-end I basically open-sourced, and then the C-end service saw no conflict, really saw no conflict. So, this is the open source part.

Then next is the company’s long-term vision—I think our goal should be AGI. Everyone’s definition of AI isn’t necessarily the same, but that doesn’t stop us from treating AGI as our goal.

From a technical roadmap perspective, actually the AGI roadmap is relatively clear. With the current generation of AI technology, if you can describe a problem very clearly, giving it complete context and instructions, it already surpasses humans. But there’s a definition here, a premise: you give it complete context, you give it complete instructions.

Then this definition is very hard to achieve. For example, in today’s meeting, we actually have a very long and strong context beforehand—everyone might have decades of context, and this is something AI doesn’t possess. So what AI can currently possess is: within a limited context, it can do better than humans.

But it still can’t replace humans. There’s still one thing missing here—continuous learning. Because humans can also learn continuously. If you hire an employee, he might spend two months getting familiar with the company environment, familiar with his work, and after two months he can get up to speed.

He can do many things. He can understand what you say—for example, if you say “call Xiao Wang over,” he knows who Xiao Wang is. But with AI, because of this context, it doesn’t have those prior two months of learning.

If you tell it “call Xiao Wang over,” you have to tell it who Xiao Wang is, what position, where he is, how to find him, what to note when finding him. You have to give AI all the context. So in this situation, AI can do it, but you can’t possibly give it all the context—it’s not realistic.

So AI can’t replace your employees. But if AI has the ability to learn continuously, like your employees, coming to the company to learn for two months, then...

[00:51:02]

...but then it could replace everyone in the world. So we’re one step away—we’re missing learning-to-learn.

AI’s development, we can understand it as a staircase. Last year’s step was CoT, chain of thought. Because we discovered that through chain-of-thought methods, we can bring intelligence to a higher level. Through it thinking by itself, we can raise this ceiling, let this AI do more things.

Then we crossed another step. This year’s step is Agent, because we discovered that using the Agent approach, even more things can be done, its capability range is larger, its intelligence ceiling is higher.

Why is it a staircase? Because each subsequent step is based on the previous foundation. Agent uses CoT, and CoT also uses the previous step, which is the language model—so no step is wasted.

So, AI’s development, the trajectory of intelligence, is traceable. This year’s step is Agent, but this Agent step—it will also be walked to completion. That is, after it solves all the problems it possibly can, it still can’t replace your employees, but it’s already reached its capability ceiling.

Just like CoT—after CoT reached its ceiling, it already surpassed the most top-tier humans, in doing math olympiad problems and writing programs it already surpassed the most top-tier humans. But it still stops there, that technology couldn’t reach AGI.

So you see, AI’s intelligence trajectory is traceable. Then after Agent, we think the problem to be solved should be continuous learning—that is, how to let the model learn continuously, rather than you giving it strong training—it should be able to do relatively long-term continuous learning like a human.

This problem is the same thing as completing tasks and so on—they’re related, they solve the same problem. We now stand at Agent, and what we can see is the next bottleneck, which is continuous learning. The next problem to solve is how to learn continuously—this is visible, relatively clear.

That is, the obstacle stuck in front, you must cross over it, and there must definitely be a way to cross over, but it needs time. After continuous learning, we might arrive at a singularity. This singularity is: when this model can learn continuously, it can already do all the things humans can do.

It could then develop its own version, could research on its own, then develop its own next version, developing more advanced AI models. So it would reach a singularity, able to achieve its own iteration.

But this singularity—it’s not really a singularity, it’s also a gradual process. This process might also be a relatively long gradual change, not a mutation. But habitually, we all think it might be a singularity.

Because a long time ago, those prophets thought there would be a singularity here, but actually it’s not a singularity, it’s a continuous process. Then after this step is complete, I think that’s when embodied intelligence arrives.

This is our conjecture—this is what we think the timeline should be: first solve learning-to-learn, then reach that intelligence singularity, the singularity that can self-iterate, and then comes embodied intelligence. After embodied intelligence, it walks into the real world, can do housework for you, can care for you in old age.

We think this is a relatively ideal roadmap, but everyone’s views differ, there’s no right or wrong. Just that we feel this roadmap is the most relaxed.

This roadmap is because for each step, the new things you need to do are very few. This roadmap lets us avoid working overtime. But if the roadmap were reversed—say, it needs to first achieve embodied intelligence—then this is very tiring to do ourselves, a very bitter task. We don’t want this kind of roadmap—we want to do it a bit more relaxed.

If we first solve continuous learning, then solve that self-iteration singularity, then solve embodied intelligence, this path is very relaxed. Because later on, you can use earlier technology to help develop later technology. After the singularity, then doing embodied intelligence—that doesn’t need humans to do, doesn’t need us to do, the model itself can produce it.

So this answers what our long-term goal is. As I told him, this is our long-term goal, which is what we call AGI.

Everyone, back to reality. Last year’s most important reality was that everyone wanted to make Chatbots, to grab C-end traffic. This year’s reality is that everyone wants to grab To B revenue, to participate in this, because if you don’t participate, you’re simply not at the table, right?

But we don’t think this is an important matter. Or rather, within our company, what we truly care about is actually those AGI roadmaps I just mentioned, and how to break through with this technology in the next step.

But one strange thing is, the thing you most want to get, you can’t get. Whereas the thing you don’t care much about, is actually fairly easy to get.

There’s a strategic advantage here—that is, in our hearts we’re thinking about AGI, what we do is AGI. So when we do those applications, do those C-end and B-end things, we simply don’t need to put too much thought into doing those things—actually very little effort suffices.

I think it’s that, standing at a technological high ground to do relatively lower-level technology, there’s this kind of dimensional-reduction strike. At least on last year’s C-end, we saw it was indeed this way. We didn’t put much effort into the C-end—at one point we even didn’t want to maintain those users, but the users just couldn’t be driven away.

Because they really couldn’t be driven away, so in the end they all remained. But slightly... Then this year’s B-end revenue, right now this growth looks relatively optimistic. I think this figure, compared to peers, should also be relatively good, I estimate. But we didn’t put much effort into doing this thing, we simply didn’t...

[01:02:45]

...do one thing—it was just done incidentally.

Doing the launch of internet intelligence, along the way, the AGI step is one I must walk, it’s a stair-step. On the road to AGI, I must pass this stair-step. So I provide these technologies to everyone through API—I didn’t do anything extra.

We’re still doing AI—this is a byproduct. I only need a few people to maintain this API, and there aren’t even customer service reps, no need for sales, nothing is needed—users come on their own.

Or rather, what’s considered the C-end users, the C-end and B-end, are all byproducts on our road to doing AGI, an intermediate output, with no conflict with my doing AGI. I’m not doing it for the sake of doing the C-end, or for the sake of doing the B-end—rather, our doing AGI is for the sake of doing AGI, and it happens to produce these things, so I take them for commercialization.

This is different from other companies. Other companies do this for the sake of doing this—for serving C-end users, or serving B-end users, they build this model. But for us, the original intention isn’t this—our original intention is still to pursue AGI.

I think to a certain degree this is a kind of dimensional-reduction strike. AGI is a bigger vision, and this vision can cohere more excellent people—it has stronger cohesion. So I have an advantage in organization, and then I use this advantage to... it’s a kind of dimensional-reduction strike.

But if you’re a commercial company, and your vision is to serve C-end users well, then it’s a different story. It has other advantages—in product, user service, traffic there will be advantages, but in technology there’s no advantage.

The currently favorable situation is that model technology is most important. You need to make the model well, others... Right, and this can explain our previous development trajectory. We truly chose AGI, and I simply didn’t think about wanting to have many users.

When we suddenly became popular last Spring Festival, that wasn’t in our script—we hadn’t thought about that thing at all. We just wanted to make this technology well. But at that time I discovered that actually our organization, compared to fully commercialized, product-goal-oriented organizations, has extra advantages in talent and organization.

This is also amazing. At that time everyone was fighting bloody battles over the C-end, and the result was led away by someone who didn’t compete. This also indeed shows the logic I mentioned earlier has merit—there indeed are talent and organizational advantages.

This talent advantage isn’t that my people are smarter than his—rather it’s how I organize this talent, how I motivate them, and then how they cooperate. This thing has an advantage. Because when you gather smart people together, it’s not that they naturally cooperate, naturally have great passion to charge toward a goal and complete it—so you need a vision.

Our earlier experience gave me the enlightenment that the AGI vision is very powerful.

OK, that’s the question. Then the next question is, what’s the importance of core interests?

I said earlier that we need to be very restrained in many aspects. But what are our core interests? Actually our core interest is only one point: our greatest core interest is maintaining the stability of the team. This is our greatest core interest, and can even be considered our only core interest.

As long as I can maintain the stability of the team, I will definitely succeed, definitely achieve AGI—it’s that simple. As long as everyone stays, we can continue, and I will definitely... basically there’s no risk. It’s just a matter of earlier or later; we’ll encounter setbacks; if we encounter setbacks and everyone stays, then I can continue again.

Money is definitely not a problem, resources aren’t a problem, the other elements are all easily obtainable. For us, there’s only one core interest, only one thing that can’t be conceded: we must maintain the stability of the team.

This is also a very big challenge we face, or rather, I think it’s the greatest risk. Of course, this risk has been substantially relieved with our recent round of financing. Because the options everyone received are relatively many, the amounts relatively large.

From a team stability perspective, as long as some of the most important employees, the oldest employees, can be stable, then others aren’t likely to leave. Even if others have fewer options, less income, they won’t leave. Because they’re not entirely here for money—everyone hopes to do this thing in an environment that can achieve AGI.

So for talent, there’s still appeal. Historically, our talent turnover has been relatively low—compared to peers, our talent turnover is always relatively low. But this is still our greatest challenge, the only challenge, you could say.

Everything else is a matter of time—everything else at most causes us to be half a year late, a year late, but won’t mean we can’t produce it. We definitely won’t lack money, definitely won’t lack resources—actually none of these are lacking.

So many of the things we do now are to maintain the stability of the team. Aside from this point, everything else I think we can do without, everything can be restrained. We’ve always been very restrained, unwilling to become adversaries with any internet giant or small company.

I hope I can empower them, or hope I can assist everyone to do this thing, hope I can help everyone do this thing. This is also, as we said before, part of our commercial meaning. The premise is that everyone doesn’t...

[01:15:12]

Under this premise, we’re very willing to assist and help anyone, even our competitors, including Alibaba, Zhipu, Moonshot, to do better. Because we don’t lose anything—we’re open source anyway, and open source also means not making boundaries too clear about how to do things—we hope you can reproduce it; if you can’t reproduce it, you speak up and I’ll tell you how to reproduce.

This is originally part of open source—it won’t be otherwise just because you’re a competitor. Of course, if you’re a partner, I’ll do more. But on the big interests there’s no conflict.

When dealing with the outside, our attitude is: we only do the AGI main line. This is what I just mentioned—GPT, CoT, Agent and so on—only the main line. The AI field is very broad, there are many things we feel aren’t on this main line—for example, 3D, video generation—I think they may not have much relation to the intelligence main line, so we won’t do them.

There are also some, for example world models—I think currently they don’t have much relation to the intelligence ceiling either, so we won’t do them. But if others do them, we’re happy to help. Whether we have time is one matter, but on interests there’s no conflict.

We also hope these AI technologies can be used in all kinds of production environments, can improve society’s production efficiency, can help all industries improve production efficiency. We’re very motivated to do this.

Whether I have time, whether I have manpower, or whether our colleagues themselves are interested—that’s another matter. But on interests there’s no conflict, and we hope to achieve this goal. And we believe this has no conflict with business whatsoever—the benefits I should obtain still aren’t reduced.

I think our having held this attitude before—actually we didn’t obtain any less because of it, we didn’t obtain any less because I open-sourced, nor because of our goodwill or my providing help to others. For example, last year’s C-end users—we still have relatively many C-end users now, and they’re still relatively solid.

Our B-end this year, I think, is also relatively optimistic. I didn’t have my commercial interests affected because of our goodwill—no effect at all, and it may even be a bonus. This looks counterintuitive, but it’s indeed this way. Or let’s think in reverse—if we violated it, it would also be inevitable—could I obtain more things? No.

There’s a question: how to understand that world models have no relation to raising the AI intelligence ceiling? I’m talking about this current stage—this is our judgment. We have seen it—this is our own AI roadmap, not the only roadmap.

From our understanding, from our judgment, the most important thing currently is to do AI training well. To do AI training well doesn’t require world models, doesn’t even require multimodality. Because if you narrow the AI training scope a bit—without multimodality, there’s just a portion of tasks you can’t do, but it doesn’t affect the algorithm’s validity.

Multimodality ultimately still needs to be done. Right now what’s important is training, then the next step to solve is the continuous learning problem, and the step after that is it asking its own questions. But there’s no world model in this roadmap, no video generation.

When video generation first came out it was very hot—it seemed this must be done, as if if you didn’t do it you weren’t an AI company. So I found it strange—actually if you just think carefully, it has no relation to the intelligence roadmap.

In fact, you’ll also find that after video generation first appeared with Sora, everyone did it, big companies and small companies. But small companies later all cut it. It has no relation to the intelligence ceiling.

But commercially, it’s a good business—commercially it’s a good business. But this has no relation to intelligence. We won’t do something because it’s a good business—we’ll only do it because it’s something on the intelligence roadmap.

Video generation—because that’s relatively clear, I use it as an example. Then world models—the meaning of world models isn’t that clear, because many things can be said to be world models.

From our judgment, world models and intelligence aren’t the most important thing at this current stage. The most important is AI training, and how to solve continuous learning after AI training. This is our company’s judgment—of course each company’s judgment differs.

What I just said—the most important issue for our company is personnel stability. From another dimension, what do we lack? What’s our gap with the US? Actually the gap is only one thing: resources.

We don’t have that many cards [GPUs]—the number of our cards is still relatively few. We currently have roughly 20,000 H-equivalent compute cards, and most of these just arrived, arrived in the last month or two, and there may be many more machines that haven’t arrived yet.

Our total compute was relatively little last year—this year we’re very aggressively expanding this compute. We currently have roughly 20,000 H-equivalent, and in the coming months we’ll have large batches of machines bought over, basically all NVIDIA.

How many cards do we need? Right now, definitely the more the better. Within the range we can afford, definitely the more cards the better—this is without question. So our current strategy is, within a reasonable price, buy as many cards as we can buy.

If, after this round of financing is spent, however many cards I can buy, I’ll buy. The pace of spending isn’t in the plan—rather, as long as the price is reasonable, however many there are, I’ll buy them all. So if I spend all the money within half a year, I think that’s a good thing. If I spend all the money within half a year, that would be too blissful, too ideal.

Actually, spending this much money is very difficult—you can’t buy that many cards, they’re very hard to buy, and prices are very high, and you can’t spend extremely high prices to buy them—you still have to ensure the price is reasonable.

[01:26:41]

If I spend all this money within half a year, that might be most ideal. Because turning money into NVIDIA cards is definitely better than putting it in the bank. Putting it in the bank, I can now maybe earn two points... but buying NVIDIA cards, ten months of social cost.

So definitely, however much you can buy, buy that much. First, after buying cards, I have lots of room afterward—through providing services or however, I can always have cash flow. If I have cash flow I can survive, so I don’t need to keep tons of money in my account.

So, we only worry about not being able to buy that many cards. If we could turn all the money into cards, then we’d unhesitatingly turn all the money into cards, and here we’re willing to pay a certain premium. That is, we’re willing to pay a certain premium to turn it into cards, because this is too worthwhile.

Even after paying the premium, actually it’s still hard to achieve this goal. So objectively speaking, if this year I could spend 20 billion [yuan], then it would count as our procurement department’s performance being super good.

Our gap with the US is mainly in resources, and the gap in people isn’t very big. In people there’s almost no gap, because it’s the same batch of people—maybe Chinese people. When Chinese people go abroad, some stay domestic, some stay abroad, some go abroad—it’s not that the smart people go abroad, no.

Actually it’s relatively random. The smartest people—maybe not that slightly more than half go abroad, but slightly less than half stay domestic. Domestic talent isn’t lacking, and our base is large—we have so many people replenishing every year.

Talent isn’t the bottleneck—resources are the biggest bottleneck. Resources first affect talent cultivation, because with little compute, our opportunities to run experiments are relatively few, so our talent overall has a gap with the US. The talent gap is essentially also because of the compute gap.

On the current largest models, we actually can’t afford to train. Even if we spent all 50 billion, we still couldn’t afford to train. Even if we could stack it up, we couldn’t afford to use it. The current largest model has about 800B active parameters; domestically, we’re still at the several-tens-of-B scale—the largest domestic model might be just several tens of B active, so that’s a difference of an order of magnitude.

If I wanted to train a model as large as [theirs], it should require 50,000 GB300s, or Huawei 950s, 200,000 cards. This is just training, not yet considering doing research. So the biggest gap between us and the US is in resources.

Our current resources, and our resources within this year and the coming months, including soon-to-arrive large resources, are only enough for us to do more experiments at the 10B-active scale. Because at the several-tens-of-B active scale, there are still many experiments to do, still many things to figure out. We should be relatively far from being able to train an 800B model—still plenty of time, don’t have that many cards.

So the difference between us and the US, I believe, is a difference in resources. We might believe that all the differences we see—including the talent difference, the model capability difference, the application difference—can all be considered as due to the difference in compute resources.

Compute resources—on one hand, cards simply can’t be bought domestically; on the other hand, our capital investment is less than the US. We’re much less at the capital investment level, and talent-based salaries account for a very low proportion within this. You see them offering salaries of a hundred million dollars, but calculating it, talent salaries still account for very little—the bulk is still compute.

This problem is currently basically unsolvable, because Huawei’s output is also limited. Because if I want to train 800B, I need 200,000 of Huawei’s newest cards, and this is just training, not yet considering doing research.

So we currently simply won’t consider competing with the US at such a large scale. Right now we lose, but we should first do it well at the scale we can afford to train and afford to use—the several-tens-of-B active scale. Then when we have more resources in the next step, then bring it up to 150B, 156B, or 250B active scale.

So between us and the US there’s a gap here that currently looks very hard to close. If you insist on training such a large model, you can train it, but you can’t do sufficient research. That is, before training, you can’t do sufficient research.

There’s another question many people care about: where will the endgame gap of all large-model competition manifest? That is, where will the final large-model gap manifest? I think this gap might ultimately not be very large.

The final gap should be three aspects: one is cost, one is time, one is user experience. Aside from these, there might not be any gap.

Cost is relatively easy to understand—providing the same service, same quality of service, at what cost you can provide it. The same service—say, like BYD’s battery, under the same technical volume, can other companies provide it at this price? I think this is a relatively difficult matter, not that easy to achieve—definitely a barrier.

So cost is definitely a difference—I think cost is probably the number-one difference. Then the second is time, when you can achieve it. A few months earlier or a few months later—it’s different.

The third might be experience—user experience still has some differences. Among these users there will still be some user stickiness and user barriers, but it may not be essential. The essence is still, first, cost; second, time. Whether you produce it first or later, whether you make the same thing faster or slower.

For long-term commercialization paths and further enriching product lines, how to price? We feel that right now the most worthwhile thing to do, most worthwhile to spend effort on, is still doing AGI. Right now we need to push AGI forward again, that is, push the intelligence floor up, push it forward.

This should, at the current stage, be more worthwhile—or a path with greater returns—than doing more product lines or considering more commercialization paths. I think for a period in the future, and for any period in the past, it’s probably been this way. That is, if we...

[01:39:46]

We spent a lot of time thinking—what’s the use of enriching our products?

What was the commercialization discussed half a year ago? It must be that I need to do advertising, then I need to do e-commerce, I need to embed e-commerce in the product, then integrate with local services and such. It’s definitely useless, because change is too fast. This product—if you’re at the front, or at this current stage of ours, and spend a lot of time on commercialization considerations and paths, the product line’s lifecycle is very short.

I think it’s not yet the time for this. This is our judgment, or at least all the previous experiences support this judgment. In the past three years, at any point when you came to talk to me about commercialization paths and product lines, it was a waste of time, because you can’t predict, can’t foresee the future. What you can foresee is very little.

Because I’m watching the time—I don’t know if everyone wants a mid-session break? Want to eat something first? Or shall we continue talking? If no one objects, I’ll continue.

On doing world-leading AGI research, and commercializing at the appropriate time—I think we’ve always been doing commercialization, I believe we’ve always been doing commercialization, just not with commercialization as the goal. We take AGI as the goal, but we’ve always been doing commercialization—that’s why we have C-end users, why we have B-end revenue.

From historical experience, this strategy is successful. Then I think the point at which we completely turn toward commercialization should be very far off. So the biggest thing is still the extension of technology, then doing the next generation of technology, more so solving the current problems. These return ratios—personally I feel that, in the currently foreseeable future, at any point focusing on products is too early.

So this is also part of our restraint. I hope these commercial opportunities can be done by others—we hope these commercial opportunities, how to use this AI, is something the entire society, all our partners do together, everyone shares this return together, rather than me wanting to swallow it all, which is impossible. And we don’t have that much energy, our organization doesn’t have that many people to do this thing.

For partners, actually this round of financing was carefully selected. As for the specific proposal of how to cooperate—but first, I think the interests are relatively aligned, that is, those most aligned with our interests, with the least hostility toward us, or those who most hope we can succeed. Not everyone hopes we can succeed, because we still harmed many others’ interests.

Company major strategy, technical and business decision processes and decision mechanisms—our company is overall built on the foundation of consensus. I don’t decide all things by myself—rather, I need to seek consensus. My authority within the company and my influence within the company are built on the foundation of consensus.

For example, if I want to do something, I definitely first look at what everyone’s consensus is, whether everyone wants to do it. Then I might have some guidance or inclination, but the effect of this guidance is limited, the effect of guidance is very limited—it’s still built on a foundation of consensus. This decision mechanism is actually a consensus-seeking mechanism—it’s not that I can push something forward; it must be consensus for me to push it through, and then I’ll push it.

Currently the main energy is basically all on the DeepSeek side. Let’s rest for five minutes. Let me write quickly. Everyone can unmute, give me some feedback. We’re fine on our end—otherwise let’s rest for five minutes first.

The final effect gap opening up between the various models should be a comprehensive one. Comparing model effects definitely has to be done at the same cost—that’s what’s meaningful. Because when you compare two cars, you also compare cars in the same price range.

A well-made model versus a poorly-made model—this difference shouldn’t be at some specific link; it should be overall.

Anthropic currently surpassing OpenAI—is this long-term? I think this isn’t long-term, this is definitely extreme. OpenAI and Google—in the future they’ll most likely still rise alternately, they should rise alternately. Actually right now Anthropic’s advantage in Code Agent isn’t that large—it’s not that it crushes OpenAI.

Our company maybe has half the people who usually feel OpenAI is better. Actually Anthropic has a first-mover advantage, but this first-mover advantage should disappear soon—it’s not an advantage it can hold long-term. All three companies are very impressive—among these three, efficiency is highest, the cost they spend, what they spend, the money they burn should be the least.

In the global AI division of labor, the role Chinese companies are very likely to play is still being the largest producer. By common reasoning, our production capacity is the largest, including chips—chips probably we have the largest production capacity, we have the most electricity, so our AI ultimately is very likely to be one of the three-body [players].

Chinese people will make this product the cheapest, and then in terms of effect—after all, foreign goods, now many goods where the Chinese-made and American versions don’t have much difference. In the future AI might be like this too, but Chinese-made AI might have a cheaper price. This cheapness might be systematically low, just like in other industries where China provides cheaper services.

When I want to do something, my habitual thinking is: what has the greatest return for me to do at this moment? If I feel doing products has the greatest return right now, I feel I’ll go do products; if I feel realizing AGI first has the greatest return right now, then I’ll go do AGI first. Obviously, I feel doing products right now doesn’t have the greatest return.

If you’re asking about the domestic-card adaptation problem—domestic cards actually have a historic opportunity right now. Because previously domestic-card adaptation had a difficulty called “poor ecosystem.” That is, after buying cards, you couldn’t use them—it didn’t have NVIDIA’s ecosystem. So NVIDIA’s moat is very strong.

But this thing is changing. NVIDIA’s CUDA moat is rapidly disintegrating. The reasons for this rapid disintegration may be three aspects. One aspect is that now with AI, after having AI, building this ecosystem is much easier than before, because AI can write code.

[01:56:36]

I can use AI to build this ecosystem, so I can build out an ecosystem exactly like NVIDIA’s. First, because there’s AI; second, there are some new technologies. For example, our company released a technology called TileLang, which is a high-level language.

Using this high-level language to write CUDA operators, we can quickly write out NVIDIA’s entire ecosystem all over again, and then combined with AI, this looks like it has no obstacles. But right now it’s not finished yet, not done yet, though this technology path looks like it has no obstacles.

There’s another point—because CUDA, NVIDIA evolved from gaming cards, so in many places, the design of gaming cards and the settings of gaming cards are of one piece. CUDA is compatible with gaming cards. Previously because AI computation was a very small field, smaller than the gaming card market, this was reasonable.

But now the compute card market is already bigger than the gaming card market, so there’s no reason for these two to still need to be coupled. The current trend is that they’ll no longer be coupled going forward. So dedicated chips—whether Huawei or NVIDIA itself—will all be dedicated chips going forward, not the previous things.

Under this background, the role of NVIDIA’s original ecosystem is greatly reduced. Because for a dedicated chip, it has no relation to CUDA—it’s not bound to CUDA. Or rather, when this chip is designed, it already considered how to build this ecosystem.

It’s a bit complex, but anyway, domestic AI chip substitution now has a historic opportunity. We believe that within the next year, we’ll be able to see one thing verified: the domestic chip ecosystem has no problem whatsoever. Previously it was thought to have problems, thought to be unusable, not easy to use, but in the future, I think within a year, we’ll be able to reverse this perception, or reverse it with facts.

Domestic AI chips’ hardware and ecosystem both have no problem—the only problem is insufficient production capacity. On this point of domestic-card adaptation, there’s no obstacle—NVIDIA can’t block it. If it were in a normal business environment where I could buy NVIDIA cards, then domestic substitution would be relatively hard; but when NVIDIA cards can’t be bought, everyone is forced into it—everyone has to do domestic chips.

Under this background, domestic-card adaptation has no obstacle whatsoever. Domestic cards building an ecosystem as good as NVIDIA’s, even better than NVIDIA’s, I think has no obstacle, but still needs time.

We currently mainly cooperate with Huawei. Huawei does their own adaptation, but we ourselves will participate in this ecosystem, will deeply participate within Huawei. Huawei’s problem is still insufficient production capacity. Like Huawei gives us roughly 16,000 cards of capacity, internet giants maybe get a hundred-something thousand, we get ten-something thousand—I think this ratio is also relatively... but this is probably just how much capacity Huawei has.

So we also can’t count on training that bigger model on Huawei later, or training a model with several hundred B parameters active—there were some saying this year. But next year, the year after, there may be opportunities.

For Huawei-card adaptation, the main work we’re doing is making its high-level-language compiler well, making TileLang well. After making TileLang well, the problem might be solved readily. This matter is relatively complex to describe, but we’re doing it, and after it’s done it’ll be much clearer to explain.

You can understand—when V3 was trained, it still used NVIDIA cards, but it already didn’t use NVIDIA’s ecosystem. V3 used NVIDIA cards but didn’t use NVIDIA’s ecosystem—rather, we first wrote a high-level compiler called TileLang, then based on TileLang’s ecosystem completed everything else, already almost not depending on NVIDIA’s ecosystem.

As long as I take this whole set of things, and redo this process on Huawei cards once, then it’s complete. I think this might be a historic mission—it can completely reverse everyone’s previous perception that domestic-card ecosystems are poor. Right now some favorable conditions of timing and location are basically all there—only time is lacking. I think within a year, there should be many people using it, or everyone understanding that this problem is essentially solved, and the rest is just the production capacity problem.

I’m relatively optimistic about domestic compute. I think on this point, NVIDIA is digging its own grave. Huawei’s supernode, Huawei’s 950 supernode, in performance and price can completely substitute for NVIDIA’s GB200, GB300. The price is definitely more expensive, but limitedly so. Fifty percent more expensive, a hundred percent more expensive—a hundred percent more doesn’t matter, two hundred percent more doesn’t matter.

For example, a hundred percent more expensive—I think it can already be considered a price-level substitute. On tasks it’s also a substitute—all the tasks GB300 can do, Huawei supernode can do, latency and everything the same. The only cost is that four Huawei cards equal one NVIDIA card, and simultaneously lagging by two years.

Four-equal-one is understandable. Lagging two years means four Huawei 950s can equal one GB300. Lagging two years means lagging two years in time. Huawei 950 supernode is this year’s Q3 or Q4 product; NVIDIA GB200 is a product from two years ago’s Q3—two years apart.

NVIDIA this year’s Q3 might already have a new generation. So the gap between us and the US on chips—I believe there won’t be a gap on ecosystem going forward, but on chips it’s fourfold plus two years.

There’s another question: will we do vertical integration upstream? I hope not. So there’s a question: will we do vertical integration into upstream applications? We hope not to—I hope others do this thing. I don’t want to gobble up everything—I only want to eat one piece, only the piece I’m best at, or the piece we ourselves consider most core, then the piece directly related to the user...

[02:08:18]

...I think probably for many of our industry partners, they care more about similar... it should be others’ meaning, I shouldn’t interpret it.

Will we build large clusters ourselves in the future? I think building large clusters ourselves is definitely necessary—we’ve always been doing this ourselves, all our clusters are self-built. But whether to develop our own chips in the future, I think depends on how big the returns here are, depends on how big the returns are.

Tesla, cultivating... this matter. Suppose you operate a power plant—you don’t necessarily need to manufacture generators, right? The power-generation equipment can be made by others, and as long as its price is reasonable, why would you make it yourself?

So, I hope not to have to make chips. I hope I can buy chips at a reasonable price, so I don’t have to make chips. I think this is very likely the case—that is, recently NVIDIA chip profits aren’t necessarily... although...

We hope to only do one piece. I think AI is a very big thing, and it doesn’t need me... I only do one piece. If we focus, and I believe the business interest here is already big enough—that is, if the AI era will produce many trillion-level companies, I think we’re one of them.

We’ve always done one small piece of it—us being one of them is already no problem, and it doesn’t need me... I’m not confidently ambitious about doing everywhere. I think it’s most likely relatively hard to change, and if you really want to do other things, you’ll still have more ideas—this will... this is our attitude.

For example, at least in the To B and To C businesses, from what’s currently visible—those who really want to make a To C closed loop, we actually do it better; those who really want to make a To B closed loop, maybe don’t do it as well as us. The more you want...

Then also, on To B I can say a few more words. The ceiling of the To B business should still be demand—under the background of the current generation of AGI/AI technology, To B demand should be limited. It’ll grow quickly, but it’s not an infinitely large thing—ultimately still constrained by demand, not compute.

Under current technical conditions, how big the revenue is ultimately still depends on demand. Because think about it, I can recover cost in ten months—I definitely would buy in bulk if there were somewhere to put it, and it’s because there isn’t that much to do. Right, demand should grow larger and larger—if with continuous technical breakthroughs, this demand will grow larger and larger.

Multimodal deployment—we’ve always been doing it. For products it’s very important; for C-end user products it’s very important. But for the intelligence ceiling, it’s a component, it’s not the main line itself.

But as a component, multimodality we’ll definitely do, and we’re doing it too. We’ll probably release related models—that is, our V4, V4’s subsequent versions will support native multimodality. But for us, multimodality—for intelligence, it’s a component—we don’t treat it as intelligence itself, its main line is the same as search.

Search is also a component, multimodality too—our understanding is it’s also a component. Scaling—we believe in scaling, definitely the larger the scale the better the effect, able to unlock more capabilities. What stops us from scaling is actually compute—it’s not that we don’t want to scale, it’s that we don’t have that much compute to do this scaling.

We haven’t touched where this ceiling is. We train models this large not because I feel a model this large is enough—rather it’s that I happen to have this many resources. I calculate based on my resources how large a model I can accept, can train—it’s calculated this way, not that this model is enough.

Currently, model returns are still very obvious. We haven’t yet had the chance to hit this scaling wall—it’s still very far from us. When Silicon Valley talks about scaling reaching its end, that’s for Silicon Valley; for Chinese people, we’re still very far from that—we simply haven’t scaled to that degree.

This scaling includes data scaling, model-scale scaling, and training cost. We’re relatively far from exploring this scaling ceiling—that is, don’t have that much compute. But we’ll also spare no effort to push this scaling ceiling.

Including after we finish financing, we’ll have more compute, and might train larger models. The shorter the time to buy, the better; if we could exhaust all energy in half a year, that would be best—actually can’t achieve it.

The next-generation model’s core capability—I think it must have continuous learning ability, only then can it be called the next-generation model. Before that, what we can do is cost, and making the effect better, making the speed faster. But for a major breakthrough, it should possess continuous learning.

Then there’s another question: why do we seem to especially care about this model’s computational efficiency? Because I’ve indeed found that not everyone cares much about this model’s efficiency. For a commercialized company, there’s no motivation to pursue model efficiency, because slightly higher model efficiency...

So it doesn’t have that much motivation to pursue very high model efficiency. So several startups—they themselves won’t... you don’t hear them say low cost is what they pursue, because that doesn’t conform to their interests. If low cost, what money do you still make? If low cost, you can’t collect much money.

On the contrary, this is part of our vision. Or rather, our vision, our colleagues care about this cost, because our colleagues are all ordinary people, they know using this costs money. They can have this empathy: others have to spend money to use it, so if others can have it cheaper, the degree others can accept will be higher.

So we have many colleagues who still hope we can make the cost lower. But if you speak from a business perspective, you actually wouldn’t think this way. From a business perspective...

[02:21:31]

From a cost or business perspective, this isn’t the first priority.

For either startups or big companies, this service cost simply isn’t high. But I think we hope it’s relatively light, I hope it’s something affordable, especially under the background of China’s compute shortage, that it’s affordable, that it can be used on domestic cards.

I think low cost is first of all a result. Our models have indeed always been moving in a lower-cost direction on model architecture—this relates to our vision. We still have many algorithmic methods, and cost can still go lower.

Another reason cost goes lower is: the lower the cost, the more I can train larger models, the more I can afford larger models. On the same compute, when compute is limited, if my computational efficiency is higher, I can afford larger models.

For big companies, they don’t necessarily think this way. For big companies, resources can be added—they can solve it by adding resources. But we’ll prioritize cost efficiency.

The value of data-type models—data is a relatively broad range, data should be almost equal to half the model. Why would I feel that if I want that, or suppose I assume AI can occupy twenty percent of GDP, and then if I want to occupy five percent within it—absolutely won’t work. Because I’ll definitely be defeated by another person—if another person says I only want to occupy one percent, then I’ll definitely be defeated by him.

If my goal is that I want to occupy five percent of AI, of all-humanity GDP—theoretically the math still works out. Look at OpenAI—their math seems to work out, theoretically no problem. But they have a problem: they’ll be defeated by another person willing to occupy only one percent.

Because another person says, I do it this well, but I only need to take one percent of global GDP—then he’d defeat him. At this point if yet another person emerges saying I only need zero-point-one percent, then he’d defeat the previous person.

Macroscopically, no matter which of these few percent it’s taken from, there’s no difference between them—they’re all the same. Those who take more will be defeated by those who take less. You don’t even actually need to take more—if your vision is to take more, you’ll be defeated by those whose vision is to take less.

Actually no one has gotten money yet—it’s just a vision. If your vision is to take more, you’ve already lost first, you’ll face greater difficulties. This is just how the world is.

OpenAI from the start felt they could truly monopolize this world, but actually they’ll encounter many many challengers. They’ll encounter challenges, so they won’t have it that easy. Challenges the US has encountered—in the future they may also encounter China’s challenge, because Chinese people are willing to take less and can provide this service.

In China, there will also be people willing to take a bit less. But ultimately there’s a balance here, because if you take too little, the company’s business logic doesn’t hold and it can’t survive. So if you take too little, you can’t survive; if you take too much, you’ll be defeated by those who take less.

So for us, we’re not about taking the most profit, or about return-maximizing pricing—rather it’s earning just a reasonable return. This is one explanation.

I believe in this thing—I’m not looking for reasons for this matter, because there’s no need to find reasons. I do it this way to begin with, and I definitely have reasons for doing it this way. These reasons may not be very conventional, but I think the company isn’t conventional to begin with.

Our company’s management actually has two lines: one line is top-down, one is bottom-up. Bottom-up is—each person does whatever they want to do themselves, no one manages them, no KPIs.

Top-down is the formal—we collectively want to do something, needing everyone in the company to cooperate. For example, if we want to release V4, then there has to be division of labor, each person does one part. That’s top-down, and this top-down we call formal.

Generally we hope this formal doesn’t take up more than half of employees’ total time—that is, formal shouldn’t exceed half. They still have half their time that’s unassigned, and they do whatever they want. This is a research scope, letting them explore on their own—according to what they feel is important, they explore, with no prerequisite requirements.

As long as the company can support it—company compute can support them doing it, or they don’t need compute, they need to do very little—then they simply don’t need to coordinate. So this is our current organizational method.

Some people think we’re top-down, some think we’re bottom-up—I think both are right. My one standard is that formal preferably shouldn’t exceed half.

We generally also don’t work overtime much. Overtime has two reasons. The first is that doing research needs a relatively relaxed environment. If you push very tightly, you can’t do research. Because it’s precisely that you need to have this interest yourself, you need to think about these problems in your spare time, so it has to be in a relatively relaxed environment for there to be a chance to explore. This is a need arising from research culture.

The second is that we’re very focused. Being very focused means the things we need to do are few. Then I don’t have that many things to do, so I don’t need to work overtime.

This is of one piece with the earlier restraint. Because I’m restrained, so many times I just don’t do it. Then the things I need to do become fewer, and the work each person gets assigned becomes less. You see many of our products are incomplete—we haven’t gone to fill them in.

This is also a culture of ours.

OK, because there are many questions, I’ve mostly scanned through them. Everyone, if you have any more questions, please ask.

“Please, investors, feel free to unmute and exchange. Let me remind you slightly—Mr. Liang talked about a lot of relatively sensitive information, please absolutely don’t spread some numbers or situations externally, including card quantities and such. Also absolutely don’t screen-record and share externally. Thank you all very much—if you have questions, please feel free to unmute and exchange.”

[02:34:59]

“Brother Yang, could you share more about the timeline for when continuous learning will bring a breakthrough? Including if continuous learning is realized, what other architectural and algorithmic innovations are needed, and other key elements?”

Few people, research is needed. Right now the whole world is researching this problem. Or rather, for investors, what investors see most now is AGENT; but for us research people, what we now see more of is learning, and how to solve this learning problem.

Actually learning may not be a single technology—it’s a problem. How to solve this problem may have many technologies—not one technology, it’s not one thing, it’ll be many things. Or rather, AGI is composed of many things, needing models, and then needing many other things.

Actually it’s also an engineering and algorithm problem. This problem is relatively specialized, but there are many methods, and many research efforts.

“Mr. Liang, thank you. Thank you very much for this opportunity today. First, I really want to respond to and express gratitude—what you said at the very beginning moved me greatly and gave us much inspiration. You mentioned that this team carries the greatest goodwill, hoping to, within this industry, push forward society and the development of human intelligence, making a little contribution.”

“And within this, carrying this sense of mission and vision. I think this is very close to the corporate culture of the company we serve, which is ‘cultivate oneself to benefit others.’ I deeply understand why you would lead the team to do open source.”

“I would vividly feel it’s like us building a banyan tree, a little-birds’-paradise ecosystem—benefiting all things without competing, but this way it’ll be accepted by everyone, all things coexisting, and ultimately it’ll be everywhere. So we’ll use this investment to likewise express our recognition, support, and respect for this mission and vision.”

“At the same time, we also hope to contribute some strength in the future of this industry, in some areas we’re good at. On this I also want to continue consulting and discussing with you. For example, in future ecosystem co-building, now after open-sourcing, how many partners, talents, and teams in this industry can relatively well reproduce our currently open-sourced models and results?”

“In the future, when taking the next step to further develop this ecosystem, in which areas do you feel more high-quality talent is needed to connect to our models and reproduce them? Or is it that GPU compute is now relatively scarce? Will everyone in the future use a kind of model-matrix approach?”

“For example, we make the large base models better and better, and partners and teams in all industries make some vertical-industry models or application models within the model matrix. How is this developing currently? In the future, step by step, two years, three years, what kind of ecosystem do you feel it’ll grow into?”

“This is the first thing I want to consult you about. Second, you just now shared with many partners many observations about AI hardware. Like AI global big companies, each might make hundred-billion-dollar-level investments, and China currently seems to have some shortcomings in hardware and compute.”

“How long do you feel this can be solved, to support our AI development, so hardware and compute shortcomings don’t hold AGI back? Do you feel this is a mission Chinese people must accomplish in the future—we can definitely make it, it’s just a matter of time and capital investment?”

“But at the same time, it may be two-sided. On one hand, the progress of models will improve model intelligence, causing single tasks or certain intelligence units’ consumption of hardware and compute to gradually decrease, no longer needing such large compute operations, because model progress will make it compute cleverly. I don’t know if I understand correctly.”

“On the other hand, hardware technical progress will make compute’s efficiency more powerful. Will this be a path where the two sides move toward each other? Right now, at this point in time, using the current 960, or H200, to make hundred-billion-dollar-level compute investments—you just mentioned you amortize it over three years, so what’s its actual lifecycle? Do you feel technical iteration is four or five years?”

“Or put bluntly, will it be that a compute center now builds a ten-thousand-card cluster with current cards, and maybe three years later it’s actually relatively not-so-advanced compute? Will there be a situation where the current stage is under construction, not enough, and three years later becomes relatively not-so-high-quality compute with redundancy? I don’t know if there’d be such a phenomenon. These two questions I consult you on, thank you.”

Thank you. The first question is the ecosystem question.

We currently feel that maybe the problem every company faces is insufficient talent. But I think this talent shortage will be phased. In the early period of every industry’s development, talent is insufficient.

Including previously making websites—when websites were first being made, there were very few people making websites, talent was very scarce. Later the internet needed server-side work, and talent was also very scarce. But this kind of talent shortage is very quickly resolved—just two or three years, because large amounts of people are cultivated.

The AI talent shortage is also phased, and we’ve already seen it substantially alleviated. Because AI people truly aren’t lacking—every company will quickly cultivate people, cultivating people is fast. So the AI industry overall—whether ecosystem, model companies, or whatever—talent isn’t lacking.

Talent scarcity is definitely a short-term phenomenon. Historically there’s never been a long-term shortage of a certain type of person. I still remember over ten years ago it was said pilots were very scarce—pilots’ cultivation cycle is very long, but it was also quickly resolved.

So everyone doesn’t need to worry about the talent-shortage problem. Also, there are currently a bit too many companies domestically doing models—still too many. The US maybe has just three; China has too many things doing base models.

Ultimately there definitely won’t need to be that many people doing base models—it’ll definitely converge. So resources are also relatively scattered, and to some degree relatively wasteful. It first...

[02:44:00]

Every company has to do the same thing, but the US has just three companies doing it, resources concentrated only in these three. China’s resources are very scattered—each company gets even fewer resources. I think this will definitely converge—definitely—but it needs a process, and ultimately it will definitely converge.

Don’t need that many companies, because right now maybe everyone feels the profit margin of doing this thing is very high, so they definitely want to do it themselves. But when they discover this thing might not have such high profit, they might stop doing it. Recently there definitely isn’t such high profit—I don’t believe there’s such high profit, because it doesn’t conform to objective law.

This means we’re at a stage: if there’s a very high profit margin, this definitely doesn’t conform to objective law. We should have a reasonable profit. So this is a current state of the industry—I think it will definitely converge.

That is, everyone doing the large-model part—among them, don’t say some company monopolizes, saying “I want to take away all the profit,” this definitely won’t work. If each company only takes a reasonable profit, then actually there’s no need for that many people to do large models. China ultimately having three or four companies competing—competition is already very sufficient, prices absolutely already enough for a price war.

Large models—maybe not to say two big companies, two small companies—that’s probably already quite enough.

As for the ecosystem, I don’t have too many thoughts. We hope to support more people, but we don’t have that much energy. We have this willingness, and there won’t be interest conflict, but whether we actually do it is another matter.

But at least there’s no interest conflict here—we hope for win-win cooperation. First, I absolutely don’t think large-model companies can take away most of the profit—this is impossible, because with so many large-model companies, the gaps aren’t that large now.

The gap is only two things: one is time, one is cost. So it won’t reach any company having windfall profits—I think it won’t reach windfall profits. Those who control cost well earn a bit more, those who control cost poorly earn a bit less—that’s simply all.

Is that an answer? Was the first question answered?

“Everyone can... you believe that in the future there will definitely be many people who can... in the future actually... everyone’s data-application cycle iterations...”

“Can you hear now? Thank you. Right, thank you for your answer, and I very deeply understand and respect your ecosystem-strategy positioning in this industry. For example, on the data side, publicly available data now—I believe model companies all already have channels to obtain it, this method should be no problem, just a matter of time and cost.”

“Then subsequently, for example, when truly reaching AGI, one setting or ideal state everyone imagines is that the model can self-iterate, self-learn—that is, train itself. So for this part, currently on data, do you feel simulated data can be put to use, or is real data’s quality highest?”

“If it still needs to come from real data, will that limit AI’s intelligence to still being within humanity’s past... this level, because it depends on the real data humans truly once had? Or can this ceiling be broken through, letting the model’s capabilities surpass all humanity’s past real [data] through simulated data, synthetic data, created data, and so on?”

I think it can be surpassed. I think there are two points of surpassing—for example, Go, AlphaGo played a move humans had never seen. That is, it’s definitely surpassing humans within a certain range.

But it may also have a ceiling, may also have limitations. But this limitation we currently can’t see. We believe, so broadly speaking, it can surpass based on knowledge humans already have, that we can already articulate.

“Then subsequently, is this by real data or simulated data? Will it work?”

There can’t help but be many methods.

“Okay, thank you. I’ve also taken up your time. I’d also like to continue consulting on the earlier AI Infra question.”

The second question was what? A bit...

“Alright, let me quickly repeat it briefly. I want to consult—for AI Infra, in the future I believe everyone is investing at the hundred-billion-dollar level in compute. So on this side, maybe we believe Chinese people in the future are mission-bound to accomplish it on hardware, that one day there might be high-efficiency compute, but maybe in the practical process, currently it’s still a constraint.”

“So in the future, will it be two directions moving toward each other? One aspect is that after model capabilities iterate, they actually go from ‘brute compute’ to ‘clever compute,’ so per-unit-model or per-task compute requirements and consumption gradually decrease marginally. On the other aspect, hardware like cards’ capability improving will make iteration speed faster and faster, single-card efficiency improve.”

“So what kind of phenomenon is this in the process? Will it be that building a ten-thousand-card cluster now, buying H200 or 960, but after two or three years, it becomes relatively not-so-high-quality compute, relatively becomes an outdated component?”

NVIDIA cards you can basically depreciate over five years. Huawei cards at most depreciate over three years. Huawei 950—this year using it is pretty good, next year using it I think is still okay, later using it I think might really be too power-hungry.

Huawei cards’ lifecycle will definitely be a bit shorter, because it’s already two years behind NVIDIA to begin with. But I think the gap isn’t that large.

If you could buy however many B200s right now, I think it’d all be worthwhile. Suppose for Tencent, if Alibaba could buy them, then look at the quantity; if they can be bought at a reasonable price, it’s definitely worthwhile. But when it’s not a matter of calculating cost—I believe they can’t be bought.

Understood. Our compute lagging is a fact. This fact is dissolved through three aspects. The first aspect is we accept model lag—we can only use smaller models than they can, and how much smaller is a training question. We have to accept a certain model lag, and smaller model sizes.

This lag has a benefit—lagging means you have more time, you have certain technologies, and this way you can use clever methods...

[02:53:59]

Thank you.

So the gap between us and the US might be lagging the US by 12 months, lagging the US maybe 12 to 18 months, or 6 to 12 months. Anyway, simply put, lagging the US by two years, then using only one-twentieth of the US’s compute to accomplish this thing.

This narrative is lagging one to two years, but using only one-twentieth of its compute. So in the future we want to rewrite this narrative—that is, we use one-nth of its compute, but shorten this time even more, shorten it to 6 months, 3 months—I think this is a goal.

And we can even surpass them in certain aspects. But under the situation where total compute still has an order-of-magnitude gap, comprehensive surpassing is unrealistic; but in certain key, trade-off-selected places, us surpassing in some areas might be possible.

Understood, thank you Mr. Liang. Full of confidence, let’s work hard together. I’ll leave time for other partners, thank you.

“Thanks for the earlier sharing. I have two quick technical questions here. In the earlier discussion of technical roadmap, you mentioned the core problem to solve at this stage is continuous learning, which is also a current research hotspot abroad, called Recursive Improvement. I’d like to ask—currently, what’s the biggest technical difficulty? From your perspective, when can this be solved? This is the first question.”

“The second question—you also mentioned first solving continuous learning, then doing intelligence. I also want to understand—what’s the technical root behind you framing it this way? Does it mean that after solving the continuous learning problem, DeepSeek will subsequently also do general intelligence? I consult you on these two questions.”

The technical question is actually a bit... [hard] to explain—its difficulty lies in that we currently still haven’t found a method that really works. The whole world still hasn’t found a good method—everyone is groping. So right now it’s still at the groping stage—that is, we don’t know who can grope to the next method to solve this problem.

Right now it’s still at the exploration stage. We have many ideas, many ideas that currently look promising, but none have been made to work yet. Right, that’s the first one.

The second is—right now internally we relatively value this narrative: training our next-version model, we hope it can help our own development. It can improve DeepSeek’s efficiency—our model first and foremost improves DeepSeek’s own work efficiency, letting it provide more help when we develop the next-version model.

Or put simply, the model we make—the first goal isn’t for everyone to find it useful, but for us ourselves to find it useful. First it’s useful for ourselves. After it’s useful for ourselves, when I develop the next-version model I’ll be faster.

Many people internally think this way: first it should be useful for ourselves, first it’s for us ourselves to use. Then this is the fastest method to realize AGI. When it’s useful for ourselves, that means others might find it useful too, but first we have to ensure we ourselves find it useful.

This narrative is a bit odd, but indeed many people think this way. Rather than users finding it useful, I hope it helps ourselves more, so we can realize AGI, will be much faster. So the logic of this narrative is: to help us realize AGI.

But first it’s to help us realize it. Realizing AGI needs this help. Right now it’s very certain—we indeed need AI to help us realize AGI. Although it doesn’t yet work autonomously, it’s still just paired with humans, but it’s already very useful.

The second question—just now it was mentioned, first solve the continuous learning problem, then enter general intelligence. Is this an expectation of yours for what follows? I want to understand what the technical root behind this is. Why must continuous learning be solved first, then enter general intelligence? Your understanding of this subsequent part.

Because solving the continuous learning problem can greatly accelerate our R&D progress. If I first solve the continuous learning problem, then the general intelligence problem is no big deal. I’d have AI’s assistance, and if AI can learn continuously, its capabilities should be very strong.

The current Agent’s capabilities are limited because it can’t learn continuously, it can’t effectively learn continuously. If continuous learning could be done first, then AI’s capabilities are very strong—it can greatly improve the efficiency of our own research.

If continuous learning is made first, general intelligence might become very easy—using it to do it becomes very easy. So I say this is a result we relatively hope to see—we’re relatively effort-saving, we’re relaxed. Otherwise, if you were to manually do general intelligence now, it’s relatively tiring, relatively bitter, a data-intensive, manpower-intensive matter, and cost-effectiveness isn’t high either.

Thanks for sharing.

A small question—the online questions, take a look at the chat group. “How much longer until AGI? Can domestic hardware catch up in this time?”

In the Zoom meeting’s chat window.

Okay, I saw this.

Huawei 950—right now Huawei gives us 16,000 cards, this should be publicly stateable. It should be an order of magnitude less than internet giants. Huawei can only give us this much, because this price isn’t cheap either.

Internet giants’ pursuit is bigger—for internet giants, they need it more. For us, we can buy some non-compliant cards. So our purpose in buying Huawei 950 is still hoping to help Huawei build this ecosystem well.

16,000 Huawei 950 cards only equal 4,000 B-series cards. So it’s not a very large quantity, the significance isn’t very big. Not enough to train a next-generation model—it’s only enough to train our current generation model, not enough for the next generation. But it can let Huawei do this well first—that’s about Huawei 950.

Then, how much longer until AGI? Can domestic hardware catch up in this time?

I think on this AI endeavor, on this AI matter, domestically in one or two years we should be able to reach roughly the same as abroad, or maybe this year we can achieve substituting for foreign models. On this AI matter, with the current approach, current paradigm, it’s not a very hard thing, so this year we should be able to achieve it. But it’s not yet AGI.

[03:05:48]

At least I feel it needs to be able to learn continuously.

Can domestic hardware catch up in this time? I think domestic hardware might need a few years. Domestically we first need to solve the ecosystem problem, because ecosystem is a confidence problem. Solve the ecosystem problem, then solve the production capacity problem—I think it should be able to be gradually solved.

I don’t quite believe that five years from now, we’ll still be stuck on the production capacity problem. Right now we’re definitely stuck on the production capacity problem—this year, next year, the year after, I think we might still be stuck on the production capacity problem, but five years later, I think maybe not necessarily—I’m still relatively optimistic.

Then the second question is thinking about future organizational structure, planning the scale of personnel. First, our previous organizational structure was very decentralized, because there was no organizational structure. But after our people expand further, these definitely need some changes made.

Right now I can only say there will need to be many changes made here, but right now it’s hard to express it all at once. Ultimately there should still be different departments—some departments we need to build a relatively rigorous hierarchical structure for. Other departments might still maintain a relatively loose, relatively flat structure.

As personnel increase, we’ll make this adjustment. It should be that we have to make this adjustment right away, because I’m already making this adjustment. If this adjustment weren’t made, many things couldn’t be pushed forward. There indeed are many departments that should have organizational structure.

Then everyone still has the question—CV is prior to which version, right? I think right now, we released this GCV4 version online—it’s still relatively rough, and in terms of capabilities there’s still a lot that needs time.

Generally, for me, a relatively comfortable release cadence is roughly releasing one version every two or three months. The last release was maybe late April, so the next release might be late June—roughly like this. If there are no surprises, each version should still be better than the last.

At the 50B-active scale, I feel ultimately we won’t have too much difference from the current wave of open source. In inference speed and effect, I feel there might not be too much difference.

But compared to their large-size model—their one unreleased model—the gap should still be relatively large. That gap should be that our active parameters can’t achieve it—that definitely has to be a larger-size model, maybe say 150B.

So 150B—I think based on our current training progress, this year, optimistically by year-end we can start training, at least next year April... the gap is still relatively large. Right, this is the gap with OCE.

“Hello, I actually have a question—you often mention this AGI, that its realization process is a gradual process, rather than there being a mutation. So can I understand this as a process with no critical point?”

It has no critical point, but it’s nonlinear. We currently relatively believe a narrative that AI can accelerate AI research, AI can accelerate AI research. That is, it’s not linear, because you can use AI to accelerate your own research, so it might be nonlinear later on.

Understood. So currently your—I understand the earlier conclusion might be that continuing to scale language models is enough, is sufficient, to reach this state.

I can only say language model scaling—I currently don’t see a ceiling. Our current intelligence level, or the intelligence level reaching the US, neither has seen a ceiling.

Understood. Because I’m very curious—you said, for the US, this 800B-active model, they can train it but they also can’t afford to use it, so they can only train it out—it’s also hard to bring out for everyone to use, because it’s indeed a bit expensive.

What I was actually curious about earlier is—humans actually having language ability is only a matter of the last hundred thousand years, and before that they maybe evolved for 3.7 billion years. But on this matter of training AI, maybe it’s possible—you say that order can be reversed—but ultimately it might still enter into so-called, not necessarily called world models, still entering into physical models or the embodied part, right? That is, after this ceiling.

Right, I think embodiment definitely still has to be entered—ultimately embodiment. So for our company, naturally, the endpoint might all be embodiment. Because for a normal person, their needs aren’t computers, right? Because normal people—their eating, drinking, entertainment, clothing, food, housing, transportation—they don’t need computers.

What they need is—so they still need embodied intelligence to solve concrete human-labor needs. If the goal is to relieve human-labor needs, then embodiment I think is unavoidable.

Understood. So speaking in stages, suppose reaching something like—not necessarily a critical point—that is, can self-evolve, can self-evolve relatively well, or approach that... the rising to that time point, I’m very curious what state of AI you’d hope lands first is...

Might be different from now. We hope it can—that is, directly, suppose there’s no embodiment, then our definition of AGI, or what we hope AGI can do—it can help me iterate the next-version model, help me iterate the next-version model.

Then after there’s embodiment, what we hope it does is likewise—let it iterate the next-version embodiment, let it make the next-version robot.

I’m still very curious about one point—because I read previous DeepSeek interviews and such, it should be that in some important direction choices and research choices, taste and intuition are very important, rather than simple engineering optimization. So if AI can self-evolve later, are these things—taste, taste, intuition—still important, or what will be important?

AI currently doesn’t lack taste and intuition—what it lacks is continuous learning ability. AI’s taste and intuition are no problem. If you have it write an article, its taste and intuition—I think there’s no problem.

There are still several questions earlier—let me look at those questions. I saw a few questions discussed on screen, but I can’t see them now on my end.

“Mr. Liang, I left a question on my screen, let me read it to you again. Actually I’d like to consult—continuous learning, as you also mentioned, many researchers also mention it’s a still-unsolved research problem. Then the coding agent, especially catching up to MILES, that is...”

[03:17:47]

“...Scaling to this Office level, MIS level is a relatively definite goal. For a still-unsolved research goal and a relatively definite scaling goal, how do you think research resources should be allocated, especially the talent resources of researchers, to achieve the best balance and effect?”

Model Office is a relatively definite goal, but model MIS—I think it’s not yet certain to say it’s definite, can only say it’s a goal. Model Office, I think, should be relatively definite.

CoT doesn’t take up resources. Doing this kind of free-form research doesn’t consume cards, needs only very few cards—what it needs is ideas; it also doesn’t consume talent resources, because it doesn’t need someone constantly doing it. It’s not a project—rather it needs many people all thinking about this problem.

So there’s no need to allocate any resources to it, because it doesn’t need resources. To train models, to make models, release models, do model-efficiency experiments—that needs resources. As I just said, the resource consumption on people and cards is both low.

So we call this “drawing lottery tickets.” The threshold is very low—anyone can go draw, but who can draw out something—this maybe I also don’t know whether it depends on talent or what.

So there’s no need for us to allocate resources here. It’s just that where we differ from other companies is that we’ll spend time discussing this problem, will think about this problem, and then treat it as an important matter.

In the company, it’s an important problem, a problem we’ll spend time thinking about, but it doesn’t need much resource to do.

Then below I’ll look at one more question: large models’ hallucination problem relatively affects user experience. The hallucination problem also has a method that can solve it, but this is a long proposition. The hallucination problem can be considered something solvable through better post-training—it’s a solvable, improvable problem.

It’s just that everyone hasn’t put much effort into it. Or rather, for me, hallucination is a problem, but we probably attribute it to a product problem. We’ll solve it, but it’s not a key problem.

Earlier there’s also a data-labeling problem. Our data annotation—this relates to our capital investment. With our capital-investment structure, we can’t support the cost of that much high-quality data annotation, because the cost is very high.

US data-annotation cost and China data-annotation cost don’t have much difference. Labeling data in China doesn’t have a cost advantage—especially on labeling high-end data there’s no cost advantage—which makes it very hard for us to invest in labeling data like the US. This path in China is very hard, because labeling data is simply too expensive—whether we outsource labeling or label ourselves, both are very tough.

So right now it’s basically walking on two legs. It’s not that we completely can’t label—rather it’s that labeling data has some low-cost, some high-cost. We label the low-cost ones first.

So you can also consider—right now half the people in our company are labeling data. Half of the core researchers, the most important people, half are labeling data. We just concentrate on labeling data. Solving AI at this current stage relies on labeling data. You just need to look—it’s about data.

“Mr. Liang, thank you. You spoke especially—we feel it very much, very good. Thank you. I’ll consult a few questions.”

“First question—you just said Chinese models are definitely stronger than US models in efficiency. You also mentioned some other aspects where in the future we might also be stronger than the US. Which aspects do you feel we’ll be stronger than the US in intelligence or other aspects?”

I think in many experience aspects, it’s possible we can do better than the US. That is, not to speak of our own experience—we feel our own product experience is pretty good—I think in user experience we won’t necessarily be worse than the US.

In products, in product capability we won’t necessarily be worse than the US. Cost should also be lower than the US, so it’s possible China will still have competitiveness.

Other aspects—if you ask whether there are structural advantages, I think maybe there aren’t. But in cost and product, these two aspects, I think there indeed are certain structural advantages.

Cost is easy to understand—it’s because they don’t need to do it, so they don’t develop this capability. They definitely don’t value this thing as much as we do. We can treat it as a very important matter, but for them, this is unimportant.

Products too—natively many companies’ product capabilities are still okay. So I think these two aspects might have structural advantages.

Okay. Second question—I consult you—you just mentioned post-training, our investment relatively high cost, like Anthropic and OpenAI both invest huge amounts. After this round of financing, do you feel we’ll increase investment in post-training?

The gap is mainly on high-quality data annotation, then mainly in AI research. We’ll definitely increase investment, but like high-quality data annotation, typically that’s not capital investment.

The bottleneck of high-quality data annotation, I think, is time—that is, needing time. Because for OpenAI, for abroad, for Anthropic, they’re all earlier, and have more capital, more cards too.

Under this situation, domestically we can consider it as having just started in the last half year, so in time, I think more time is needed. This has little relation to capital investment, because even without more capital investment, the original capital is enough for it to expand at the fastest speed.

But this speed has a ceiling—the bottleneck isn’t that I can immediately have more people, it’s not stuck on money, nor stuck on cards. But it’s indeed in a rapid-expansion process.

So we feel within a year, the high-quality data problem done relatively well—I think domestically it should be expectable. I think the outlook might not be that cold, but it indeed needs time.

Thank you. Then the third question—we’ve seen Anthropic use their own model to make their own products, launching many vertical finance, legal...

[03:29:17]

“...even in the future wanting to move toward the medical direction. Do you feel that at some stage we’ll also consider this kind of vertical application?”

I haven’t yet thought clearly about what our domestic business model will be in the future, or what the smoothest path is—we haven’t reached that stage. Domestic and foreign situations aren’t necessarily the same—what it’ll be like domestically at that time, I think is relatively hard to judge right now.

Domestically, looking at the current situation, I think the most reasonable approach should be to go all-out on general Agent, and other Agents’ priority should be lower, including finance, doctor and such Agents. Do Coding first, because Coding Agent can accomplish a lot, and there are many vertical Agents. At the current stage, what we feel is most important should still be Coding Agent.

Explained very clearly, thank you. There’s one more question we’d like to consult you on. Actually we do DeepSeek, and we also admire you greatly—you’ve always done DeepSeek in a very pure scientific-research manner.

But now this industry has indeed also walked onto the capital markets, walked onto the capitalization path. You’re also a very responsible person—whether to partners or investors, you’re all very responsible. So, between the direction of pure scientific research, purely doing AGI, and the capital markets, the balance—you’ll definitely still go to the capital markets in the future, definitely still have public-shares channels—how do you feel about balancing this going forward? How do you consider this?

I currently feel it should be achievable—that is, both. Suppose this year I can have several hundred million dollars of B-end revenue, plus we have C-end users, then this itself already has a certain commercial foundation. With next year’s B-end revenue, if this demand can grow further, the company isn’t far from net profit—it might already be net profit.

It might already not be a pure cash-burning stage, so I feel the actions we can take, can operate afterward should still be relatively large. Or rather, in the worst case, selling API might already be able to support a public company. That is, if there’s no new technical progress later, if our technology is frozen here, then ultimately we’d go all-out selling API, do these services well—I think that’s enough too.

So I think there’s still confidence—it’s indeed not that hard. Because it’s indeed in a high-leverage place, and in a very fast-developing field, it might just not be that hard. I can only say we hope to have bigger dreams, but we also have a floor of performance we can bring out.

Thank you, well spoken. Questions are done, thank you.

“Thank you for sharing. Some questions mentioned earlier, I’d like to ask you again. Because I think DeepSeek’s biggest difference from other companies is that our organization is different from others, or our organizational form is different.”

“But the organizational form relates to our goal, and maybe also needs to consider the organization’s own boundaries and efficiency. I don’t know—from a macro perspective, does our organizational form have a good learning object? Historically maybe Bell Labs, or what kind of form might be relatively ideal? Or do we ourselves feel there isn’t an ideal one, and it more needs us to gradually explore?”

Maybe in a new era, we can only rely on ourselves to explore. Because our organizational form is definitely different from the several US companies, right? The three companies are also different from each other, but they at least explore from the angle of a commercial company.

I mainly want to ask again about this organization question.

First, we have no imitation object. Every step is us proceeding from the actual situation, seeking truth from facts, making decisions according to the actual situation, finding out how we should do it. So it’s a product of the times, or a reflection of the actual situation—it’s not a result of imitation.

That is, in this situation, the optimal solution I feel might just be this way, or I myself believe the path we chose ourselves is this way. Every step we’ve definitely thought through, definitely chosen—anyway the choice result is chosen this way. It’s not that because we saw someone choose this way, we chose this way—rather it’s because we analyzed the pros and cons and chose this way.

So in the future it might be the same—we haven’t imitated anyone. I think we’re still different from Bell Labs, because it explicitly didn’t need commercialization, because it was a... But we explicitly need commercialization. We ultimately still need to survive—we’re a company after all, the government won’t give me a single cent.

So we can have a very grand mission, but at bottom we’re a company, we need to consider how to survive. So B-end is definitely important for us, because maybe in the future we’ll need it to survive. It’s just that right now it’s unimportant, because right now it’s a cost line.

So I think this is still different—still different from Bell Labs. We’re essentially still a company. Historically there have also been many companies that have pursuits beyond profit, but you can’t say that because it has pursuits beyond profit, it’s not a company.

Many companies are great precisely because they have a pursuit beyond profit. That pursuit ultimately not only didn’t affect its commercialization, but rather let it commercialize better. We’re essentially still a company—it’s just that in considering which money to earn, when to earn money, how much to earn, what to earn by, we have trade-offs.

“Mr. Liang, I have two small questions to quickly consult you on. One is—you actually mentioned MILOS earlier, that it might not be a very definite goal, but you’ll definitely work toward this direction. Then you also mentioned active parameters might for example...”

[03:41:02]

“...the next generation might be around 150 to 250B. In this situation, do you feel 150 to 250B benchmarks against O4.7, or might benchmark against something else? This is the first question.”

“Second question—you also mentioned earlier that we now use some compilation languages other than CUDA on the whole inference side. I understand originally we were based on NVIDIA’s ecosystem, maybe using PTX and such more. Now using more of things like the TileLang you just mentioned—will it substantially reduce some of our efficiency on inference, or reduce efficiency short-term?”

“I don’t know how you’d view the efficiency loss brought by this change in compilation language, or whether long-term it’s actually a supplementable state, improving efficiency?”

It improves efficiency, substantially improves efficiency.

“OK, so there’s no negative impact at all?”

Right, it substantially improves efficiency. So this is an opportunity. It’s equivalent to—previously you couldn’t leave CUDA’s ecosystem, now we can abandon its ecosystem and use a simpler method, just use TileLang. It’s a high-level language, writing those programs is also very fast, the code you need to write is very small now, I can rewrite it all over again.

Understood. So both of these are actually one—like the relatively big opportunity brought by AI you mentioned, not a shortcoming that maybe needs to be short-term made up for.

Right, it’s a big opportunity of technological development. It’s not AI’s opportunity—because we also have a project, we’re using AI to write TileLang.

Is that faster?

Right now all TileLang is human-written, but it’s already much faster than originally writing CUDA.

Understood. So even for the hardware-underlying execution efficiency, you feel there’s also no impact?

Losing 1% to 2%, I think is acceptable.

OK, understood, got it. Alright, thank you, very clear.

Does anyone else have any questions?

No questions, let’s wrap up here today.

Okay, thank you.

Thank you all very much for your time, thank you Mr. Liang, thank you.

Bye-bye.

But that is a story for another day. Today I’m talking about the upcoming Politburo meeting. By convention, the Politburo convenes in late July to evaluate the economy and set the agenda for the second half of the year. Of the Politburo’s regular meetings, the April, July, and December sessions are the ones that typically centre on the economy, and the July meeting, falling at the year’s midpoint, has long been the best window into where macro policy is headed in H2.

Two things make the meeting important. First, 2026 is the opening year of the 15th Five-Year Plan, so the mid-year tone-setting will be about not just this year’s growth target but also how the entire five-year plan gets off the ground. Second, the economy is pulling apart. In the first half, new growth drivers such as high-end manufacturing, the digital economy, and modern services accounted for over 40% of growth, with value-added in high-tech manufacturing up 13.3%. Meanwhile, retail sales grew just 1.3% year-on-year and fixed-asset investment fell 5.7%.

That means policymakers must give the new drivers room to grow without resorting to the kind of indiscriminate, flood-like stimulus that would derail the structural transition they want, while also not standing by as the old economy drags the overall GDP number down.

This is why the July Politburo meeting will be a window for reading second-half policy. And its content is not unpredictable; some direction can already be seen from the Premier’s economic symposium and from statements at official press conferences.

In short, I do not expect a large short-term stimulus package at the end of the month, but there may be structural policies on fiscal support, employment, and services consumption.

Subscribe now

1. The official read on the economy remains positive

To judge how strong policy will be, first look at how policymakers define the current situation. Judging from recent statements, the leadership’s view of overall growth and the full-year target remains upbeat, and their attention regarding Q2 leans more toward cultivating new growth drivers (培育新动能), which is basically tech sector growth and industrial upgrading. At the economic symposium on July 13, Premier Li Qiang’s judgment was

We must grasp the economic situation as a whole, and in particular see that China’s economy is fully shifting onto a track of high-quality development, and that the underlying support for long-term improvement is more solid.
要注重从整体上把握经济形势，特别是要看到我国经济正全面转入高质量发展轨道，长期向好的内在支撑更加坚实。

He also stressed “maintaining strategic resolve on high-quality development” (保持高质量发展战略定力). From this wording, I see no trace of a large short-term stimulus.

Similarly, at the press conference on first-half data, the National Bureau of Statistics used four characters to describe the economy: 稳 (stable), 韧 (resilient), 新 (new), 优 (improving). “稳” refers to GDP growth staying within the 4.5%–5% range. “韧” is language often seen in last year’s Politburo statements during the US-China trade war, stressing resilience against external shocks. “新” refers to emerging industries, represented by AI and electric vehicles. “优” refers to the share of manufacturing and the growth of corporate profits. Responding to the 4.7% first-half GDP growth, the NBS said

For an economy of China’s enormous scale, 4.7% growth is hard-won. Growth did slow in the second quarter, but the fundamentals — steady operation, and a shift toward the new and the better — remain unchanged.

对于我国这样超大规模的经济体，实现4.7%的增长难能可贵。尽管二季度增速有所回落，但是经济稳定运行、向新向优的基本面并没有变化。

As for the 4.3% growth in Q2, the statistics authorities prefer to attribute it to certain industries being affected by short-term and external factors, while stressing that monthly trends still show positive signs. This explanation means policymakers, for now, do not see the Q2 slowdown as a trend decline, so the perceived need to launch a large stimulus immediately is limited. Overall, although many academics and advisers believe the economy is cooling and have raised the problem of divergence across sectors (I recommend Li Daokui’s article, translated by Yuzhe He, and Liu Shijin’s article, which I introduced last week), from policymakers’ perspective, positive factors still dominate.

2. The trade-in policy (以旧换新) is hitting diminishing returns

Until now, the main policy tool for boosting consumption has been goods consumption, centered on the trade-in policy (以旧换新). But this policy faces diminishing marginal returns. In raw terms, sales of trade-in-related goods exceeded 2.6 trillion yuan in 2025, adding more than 1 percentage point to retail sales growth.

However, Li Xunlei, an economist and policy advisor, pointed out in a related study that the boost was strong in the policy’s first year but weakened in the second, likely because consumers who had already replaced their durable goods would not replace them again soon. For high-priced items like cars, subsidy recipients may also largely overlap with people who already planned to buy, meaning part of the subsidy covered existing demand rather than creating new consumption.

As a result, this year’s pro-consumption policy is gradually shifting from durable-goods subsidies and goods consumption toward services consumption and toward a focus on the supply side.

Premier’s symposium remarks reaffirmed that from another angle,

Implement well the policies for expanding and upgrading the services sector, enlarge the supply of quality services, and speed up the cultivation of new growth points in consumption.

抓好服务业扩能提质政策落实，扩大优质服务供给，加快培育消费新增长点。

3. New language from the PBoC, but no easing signal

What I find worth noting is that the central bank’s Q2 regular meeting, when analysing domestic and international conditions, mentioned two kinds of divergence: “divergence in the performance of major economies” (主要经济体经济表现分化) and “structural divergence” (结构分化). The PBoC didn’t specify what it meant by the latter. The most likely reading is a K-shaped split between emerging tech and traditional industry, as well as production versus demand.

Judging from the policy advice circulating among government advisers, and from documents like the 15th Five-Year Plan for the Employment-First Strategy (实施就业优先战略”十五五”规划), Beijing’s concern currently centres on job displacement. The government’s response, at this stage, centres on assessment, retraining for job transitions, and the development of new occupations, rather than legislating to slow the displacement itself. Given US-China competition in AI and the leadership’s recent emphasis on tech breakthroughs in the new industrial revolution, I don’t see China pulling back on AI investment or passing laws to halt automation. So if the July Politburo does take up structural divergence, short-term policy will more likely land on stabilising employment and vocational training.

Monetary policy keeps the earlier “moderately loose” (适度宽松) framing and continues to stress counter-cyclical and cross-cyclical adjustment, but there was no new language clearly signalling another round of RRR cuts or interest rate cuts (降准降息).

Some other things

Just as I finished writing this piece, the Ministry of Finance released the fiscal data for the first half of the year. There are two interesting numbers worth paying attention to.

The first is stamp duty on securities trading: 154.9 billion yuan, up 97.3% year-on-year. Since this tax is levied solely on stock turnover, a near-doubling means trading activity in H1 roughly doubled as well, a clear sign that households’ risk appetite (mine included) has warmed up considerably.

The second is revenue from the sale of state land-use rights, 977.8 billion yuan, down 31.5% year-on-year. That’s less than one trillion for the half-year, for a revenue stream that, at its peak, brought in 8.7 trillion yuan annually. The tightening of local governments’ purse strings shows up on the spending side too: expenditure under government-managed funds fell 16.4% year-on-year. Special-purpose bonds and special treasury bonds have plugged part of the gap, but the contraction in local government investment is unmistakable.

For policy, the widening decline in land sales arguably leaves room for the Politburo to strengthen its language on property. That said, I’ve also heard an interesting counter-reading, which is the retreat of land finance (土地财政) may not necessarily be in need of rescue, but actually a catalyst for the next round of China’s fiscal transformation; they believe it’s a shift from locally driven land finance to centrally driven equity finance (股权财政). And since land revenue sits in local hands while the capital market answers to Beijing, this switch would also mean a reshuffling of fiscal power between the centre and the localities.

Britain, the dominant power of the steam age, was eventually surpassed by the United States during the Second Industrial Revolution. The United States’ victory in the Cold War-era technological competition likewise reflected its dominance in information technology. Under such a background, China’s approach to the AI-driven industrial revolution is shaped by a strong sense of urgency, perhaps even FOMO.

China has emphasized science and technology as productive forces since the Deng Xiaoping era, especially through the slogan that “science and technology are the primary productive forces.”科技是第一生产力, but this was largely the mindset of a latecomer trying to catch up after falling behind. In the 1990s, the internet revolution was still led mainly by the United States in terms of technology. China got on board and built globally competitive companies, but much of its innovation remained concentrated at the application layer.

AI may be different. China’s technological gap with the United States has narrowed to roughly six months. (Although given the speed of AI development, six months can still be a substantial distance) More importantly, the technological trajectory has not yet been fully blocked. China still sees an opportunity to compete, and potentially to catch up.

This sense of urgency is also evident in Xi’s speeches at the National Science and Technology Awards Conference, the Academicians’ Conference of the Chinese Academy of Sciences and the Chinese Academy of Engineering, and the 11th National Congress of the China Association for Science and Technology. (yes, three meetings in one) Throughout that speech, he focused on the bottlenecks and weaknesses in China’s scientific research system. One line stands out in particular:

The situation urges us forward and compels us to act. We must seize the historic opportunity, meet the challenges of the times, and accelerate the achievement of high-level scientific and technological self-reliance and self-strengthening.

形势催人，也逼人。我们必须抓住历史机遇，迎接时代挑战，加快推进高水平科技自立自强，

My reading is that this was a direct message to experts across different fields: the window of opportunity will not remain open indefinitely.

Returning to the WAIC speech itself, its tone was relatively restrained. Its main ideas can be summarized in four points.

First, open source, openness, and cooperation. Xi encouraged open-source development and international cooperation, while arguing that AI is moving from the “digital world” into the “physical world.” This both reflects China’s comparative advantage in open-source foundation models and helps offset barriers created by closed-source ones. It also suggests that policy attention may increasingly shift toward embodied AI, intelligent manufacturing, and other real-economy applications.

Second, safety and controllability. The statement that AI must “always remain under human control” is broadly consistent with international consensus. (unless someone really wants a Ghost in the Shell-style society). But the second part of the argument is more politically pointed: China opposes the “overgeneralization” of national security concepts in AI and rejects the idea that one country’s security should be placed above others'. This is clearly an implicit criticism of U.S.-led technology restrictions and export controls.

Third, the relatively softer language of “mutual learning among civilizations.” In a bit of substance, however, this is also about who gets to shape the values embedded in AI systems. Whether AI serves a particular set of values defined by a small group of countries, or different societies should have a role in defining AI governance and norms.

Fourth, improving global governance. Xi emphasized AI capacity-building for countries in the Global South, rather than allowing a small circle of technologically advanced countries, like those involved in the G7 Hiroshima AI Process, to set the rules for everyone else.

I know many will say that China is just saying fancy words. But does anyone really expect China to frame its position the same way as the US does? Just putting its own interests above all others?

World Artificial Intelligence Cooperation Organization(WAICO)

The institutional development that took place the day before the conference is also significant. Representatives from 29 countries signed the Agreement on the Establishment of the World Artificial Intelligence Cooperation Organization in Shanghai, formally creating an intergovernmental organization: the World Artificial Intelligence Cooperation Organization, or WAICO. Its headquarters will be located in Shanghai.

The founding members are mainly countries from the Global South, with no G7 members, core EU countries, or Five Eyes members involved. This suggests the emergence of a parallel governance framework outside the existing Western-led system—one promoted by China and centered on developing countries. On this point, I also want to recommend reading the excellent piece by

The Newsletter by George Chen

AI Diplomacy Divides the World: U.S. Rallies Allies with Pax Silica, China Launches WAICO for the Global South

“The Chinese often say, ‘A single string cannot make music, and a lone tree does not make a forest.’ The development of artificial intelligence should not be a solo performance by one country, but rather a sympho…

Read more

10 days ago · 5 likes · George Chen

WAICO’s core mission appears to include four main areas:

1. Building a cross-border AI supply-and-demand platform, allowing countries with AI capabilities to connect directly with countries that need technology, infrastructure, or applications.

2. Carrying out capacity-building and inclusive-development programs to narrow the global AI divide.

3. Promoting coordinated governance on AI safety, cross-border data flows, and ethics, while jointly developing internationally applicable technical standards for large language models, humanoid robots, and related technologies.

4. Supporting open source and the sharing of scientific resources in order to lower the barriers to AI research and development for smaller and less-developed countries.

For the Global South, the most immediate problem is often not AI safety, but the lack of basic capacity to develop and deploy AI in the first place. Without such capacity, these countries risk falling even further behind in the next technological revolution.

Institutionally, WAICO adopts a structure consisting of a General Assembly of all members, a council, and a permanent secretariat. It explicitly rejects a single-country veto, highlighting a principle of “development first” rather than governance focused solely on regulation and risk control.

WAICO could give China greater influence over technical standards and create a stronger ecosystem pull for its AI industry. At the same time, it will need to address questions of funding, long-term sustainability, and the balance between promoting development and managing risks.

Overall, the agreement suggests that global AI governance is entering a period of dual-track development. The U.S. and its partners emphasize values alignment, security, and export controls; China emphasizes development, capacity-building, open cooperation, and the broader diffusion of technology.

Subscribe now

Below are Xi’s full remarks :

Working Together to Build a Fair and Equitable Global Governance System for Artificial Intelligence

Keynote Address at the Opening Ceremony of the 2026 World Artificial Intelligence Conference and High-Level Meeting on Global AI Governance

(July 17, 2026, Shanghai)

Xi Jinping, President of the People’s Republic of China

Distinguished Colleagues and Guests,
Ladies and Gentlemen, Friends:

Seventy years ago, a group of young scholars first put forward the concept of artificial intelligence at the Dartmouth Conference in New Hampshire, United States. Over the past seven decades, AI scientists and developers around the world have continued to explore the unknown, press forward through twists and turns, and achieve breakthroughs through perseverance. Seventy years on, facing a new wave of vigorous AI development, we have gathered here on the banks of the Huangpu River to discuss together how to promote the development of global AI in a direction that is oriented toward good and benefits humanity—a gathering of great significance. On behalf of the Chinese government and the Chinese people, I extend a warm welcome to all of you.

Looking back at history, the invention of the steam engine ushered in industrial civilization, the spread of electricity illuminated modern society, and the birth of the Internet connected the entire world. Every technological revolution has profoundly reshaped the way humanity produces and lives, propelling great leaps in economic and social development.

Today, the world is undergoing accelerating changes unseen in a century, a new round of scientific and technological revolution and industrial transformation is achieving breakthroughs at a faster pace, and global AI innovation has entered a period of unprecedented dynamism. Intelligent technologies—featuring the intelligent connection of all things, human-machine collaboration, cross-domain integration, and co-creation and sharing—are releasing enormous energy in combination, presenting both tremendous opportunities and governance challenges. Humanity must now confront the questions of our times: When machines begin to think, how should humans coexist with them? When algorithms take part in decision-making, how can safety be ensured? When technology challenges ethics, how can governance keep pace? When the divide keeps widening, how can inclusive benefits be achieved? These questions require the international community to reflect deeply and answer together.

China believes that all countries should uphold a people-centered vision and the principle of AI for good, making artificial intelligence an important driving force for promoting common prosperity and safeguarding common security, and work together to build a fair and equitable global AI governance system. To this end, I wish to propose four points.

First, upholding openness and win-win cooperation to drive innovation and development. Artificial intelligence is a new engine of world economic growth and an accelerator for the transition from old to new drivers of growth, and it is moving from the “digital world” into the “physical world.” We should seize this rare historic opportunity, encourage open-source practices and openness, cooperation, and sharing, and comprehensively promote AI-related scientific and technological innovation, industrial development, and application scenarios. We should work in concert to advance the transformation and upgrading of traditional industries, foster and strengthen emerging industries, and make forward-looking plans for future industries, so that AI can empower every industry and sector.

Second, strengthening risk awareness to ensure safety and controllability. Artificial intelligence should become a trustworthy tool for humanity. We must attach great importance to the various inherent and derivative risks arising from AI, and promote the development of systems for laws and regulations, technical monitoring, risk early warning, and emergency response, so as to build a firm foundation of safety, guard against misuse and malicious use, and ensure that AI always remains under human control. At the same time, we should jointly oppose practices in the AI field that overstretch the concept of national security or place one’s own security above the security of other countries.

Third, encouraging inclusiveness to promote mutual learning among civilizations. The development and application of AI should not erode or damage the diversity of world civilizations or the cultural distinctiveness of individual countries. We should shape the values of AI with the common values of humanity, make good use of AI technology to enhance understanding and tolerance among different civilizations, promote exchanges and mutual learning among civilizations, and carefully cultivate a flourishing garden of civilizations in which each appreciates its own beauty and all beauty is shared.

Fourth, advocating solidarity and cooperation to improve global governance. Artificial intelligence is the crystallization of the wisdom of all humanity and a precious asset shared by all. We should practice true multilateralism, give full play to the important role of the United Nations, and strengthen alignment and coordination on AI development strategies, governance rules, and technical standards, so as to form at an early date a global governance framework based on broad consensus and enable this frontier technology to better benefit human society. We should carry out extensive international cooperation to help countries of the Global South strengthen capacity building, bridge the digital and intelligence divide, promote sustainable development, and avoid creating new historical injustices in the field of AI.

Ladies and Gentlemen, Friends!

This year marks the beginning of China’s 15th Five-Year Plan. The 15th Five-Year Plan draws the blueprint for China’s economic and social development over the next five years, and also offers a list of opportunities to the international community. In recent years, China has actively embraced AI development, adhering to the combination of an effective market and a capable government, strengthening AI scientific and technological innovation, and vigorously advancing the “AI Plus” initiative, cultivating a healthy ecosystem in which all kinds of actors coexist and prosper together. The scale of the core industries of the intelligent economy has already exceeded one trillion RMB. All manner of intelligent devices have entered millions of households, delivering tangible benefits to people’s lives, and “Intelligent Manufacturing in China” has become another shining hallmark of Chinese modernization.

At the same time, China has always placed equal emphasis on development and security, grasping the trends and laws of AI development, and continuously improving relevant laws and regulations, policies and institutions, application norms, and ethical guidelines, to ensure that AI is safe, reliable, and controllable—so that this thousand-li steed of artificial intelligence can gallop both fast and steady.

As a responsible major country, China has always been committed to serving as a provider of international public goods in the field of AI. Since I put forward the Global AI Governance Initiative, China has promoted the adoption by consensus of the UN General Assembly resolution on international cooperation for AI capacity building, released the AI Capacity-Building Action Plan for Good and for All and the “AI Plus” International Cooperation Initiative, and proposed the establishment of the World AI Cooperation Organization—contributing a steady stream of Chinese solutions.

As the Chinese often say, a single string cannot make music, and a single tree cannot make a forest. The development of AI should not be a solo performed by any one country, but a symphony of global cooperation. Through the joint efforts of all parties, the World AI Cooperation Organization has come into being in Shanghai—a vision conceived a year ago has now been transformed into a tangible outcome. This is a major step by China in responding to the call of the Global South and rallying the international community to actively advance AI development and governance, and it will stand as an important milestone in the history of AI development.

To further support global AI development and advance global AI capacity building, I hereby announce that over the next five years, China will provide 5,000 places in specialized AI training programs for developing countries; build international AI application cooperation centers for ASEAN, the Arab League, the African Union, CELAC, the Shanghai Cooperation Organization, and the BRICS countries; and promote the deployment of “Mazu,” the intelligent meteorological early-warning solution, in 30 countries—watching over the lights of countless homes and safeguarding calm across the four seas.

Ladies and Gentlemen, Friends!

As the ancients said, “The wise adapt to the times; the knowledgeable act according to circumstances.” The more rapidly AI technology advances, the more firmly we must anchor its direction toward good and the benefit of humanity, the more precisely we must calibrate the scale of regulation and governance, and the more promptly we must improve measures to prevent loss of control. We must always guide the development of AI with human wisdom and international consensus, so that it truly becomes a powerful positive force for enhancing the well-being of all humanity and advancing the progress of human civilization.

China stands ready, with a more open posture, more practical actions, and a more far-sighted vision, to work with all parties to grasp and address the opportunities and challenges of AI development, and to join hands in creating an even brighter future for human society!

Thank you.

To my mind, however, what made the meeting interesting was the guest list. Alongside the usual macroeconomists sat Yu Donglai, a private entrepreneur who runs a department store and retail business. Unlike so many of his peers, who seem to have faithfully inherited the “Protestant ethic” of Max Weber’s telling and treat moneymaking as a calling in itself (and I’m not planning to point out a name), Yu pays his employees wages far above the local average and shares the company’s profits with the entire workforce. At a moment when weak household consumption has become the economy’s problem number one, giving such a businessman a seat at the table is a policy signal in itself.

Liu Shijin has been making a similar argument from a different angle. A former Deputy Director of the Development Research Center of the State Council (DRC), a top policy research and advisory body serving China’s central government, Liu is long regarded as one of the country’s most influential economists. In recent years, he has emerged as a prominent voice urging Beijing to shift its policy focus toward boosting household consumption, narrowing income gaps, and raising rural residents' pensions.

Liu believes that under a “tournament-style” competition model, local governments competed intensely around GDP targets. Combined with the export-oriented economies of coastal provinces, this sustained rapid investment growth over a long period under the earlier conditions of a shortage economy marked by insufficient supply. The structure of high savings and low consumption also provided, in objective terms, a funding source for large-scale investment in a catch-up economy, while during the period of medium to high growth, the rapid expansion of real estate and infrastructure masked the structural deviation of a persistently low consumption share of GDP.

As real estate slumped and local infrastructure spending decelerated, the long-hidden consumption shortfall was laid bare. With excessive productive investment driving capacity utilization and returns on investment steadily lower, the mechanisms embedded in this model, including excessive income disparities among households and low dividend payouts by enterprises, have suppressed society’s overall capacity to consume. The contraction of end demand is now, in turn, eroding the foundations of investment growth, and the old development model has become unsustainable.

On policy, Liu argues that the breakthrough should focus on reforming the savings and consumption formation mechanism to convert excess savings into effective consumption, and he proposes three quantitative targets. First, within roughly three years, the economy-wide savings rate should be brought below 40% and the corporate savings rate below 20%. The direct route is to raise corporate dividend payout ratios, including transferring around RMB 20 trillion of listed state capital, out of a total A-share market value of roughly RMB 120 trillion, into the social security fund, while also encouraging private enterprises to increase dividends. Second, tax reform focused on personal income and property taxes should be advanced steadily to bring the Gini coefficient gradually down to 0.4 or below, thereby raising society’s average propensity to consume and lowering the household savings rate. Third, the center of gravity of macroeconomic management and government functions should shift from the supply side to the consumption side, with more than half of macro stimulus spending devoted to building the social security system for low- and middle-income groups. Priorities include resolving the housing difficulties of migrant workers and other new urban residents, and markedly raising pension levels under the basic pension scheme for urban and rural residents, which covers 550 million people. In his view, this is a livelihood issue, but even more, it is an urgent requirement to restore investment and stabilize macroeconomic growth today.

Below is the English version I made with the help of AI:

Liu Shijin: Deepening Reform of the Savings–Consumption Formation Mechanism to Reverse the “Strong Supply, Weak Demand” Imbalance

The imbalance between strong supply and weak demand is a major challenge facing China’s efforts to stabilize growth at the current stage. The 2025 Central Economic Work Conference stated explicitly that the contradiction between strong domestic supply and weak demand is pronounced, and that insufficient effective demand is the salient sticking point in the current economy. This imbalance stems from a lopsided growth model underpinned by a series of institutional and structural factors. To reverse it as quickly as possible, attention and the point of breakthrough should be directed toward deepening reform of the savings–consumption formation mechanism, substantially lowering the savings rate, and correspondingly expanding end demand.

Share

I. Manifestations of “Strong Supply, Weak Demand” and Its Deeper Causes

“Strong supply, weak demand” refers to a set of seemingly contradictory phenomena frequently observed in reality—often described as a K-shaped pattern of development. On one hand, innovation is thriving, with major advances in artificial intelligence, robotics, and green technology; on the other, economic growth remains under sustained pressure, with nominal growth running below real growth. Exports have grown extraordinarily—the goods trade surplus reached a record of nearly USD 1.2 trillion in 2025—yet insufficient domestic demand continues to worsen: investment contracted for the full year of 2025, was down 4.1% year-on-year cumulatively in the first five months of 2026, and total retail sales of consumer goods fell 0.6% year-on-year in May. China’s manufacturing sector and industrial chains enjoy global competitive advantages, yet many industries suffer from severe overcapacity while the needs of certain groups go unmet.

This is not a short-term fluctuation. It stems from a series of institutional and structural factors: a GDP-oriented mechanism of competition among local governments; the dominance of state-owned enterprises—with a high share of state capital—in foundational sectors and strategic industries; a macro policy orientation that stabilizes growth chiefly by driving investment, especially infrastructure investment; and a manufacturing-centered, export-oriented economy. Together these have produced a lopsided growth model characterized by high savings, high investment, and high exports—one that prioritizes production over consumption.

The initial condition of reform and opening up was a shortage economy. During the long period of insufficient supply, this growth model was well suited to circumstances and indeed advantageous. Even as China’s economy shifted from high-speed to medium-speed growth, and from supply constraints to demand constraints, the model could keep running—displaying strong inertia and resilience—so long as room for investment remained. But its effective operation has a boundary: the point at which investment can no longer be pushed further and turns negative. That point has now been reached.

II. Insufficient End Demand as the Direct Cause of Persistently Weak Prices and Slowing Growth

The relationship between this imbalance and economic growth can be analyzed using a framework of the “height” and “breadth” of growth. Height refers to the expansion of the growth possibility frontier supported by productivity gains, driven by technological innovation, improved management, institutional reform, and opening up. Breadth refers to the extent to which different segments of society—say, divided into ten income groups—generate effective demand for existing productive capacity. Height determines the growth rate that is attainable; breadth determines the growth rate actually achieved.

Strong supply raises the height of growth, while weak demand means the breadth of growth has not expanded commensurately. Gains in height cannot substitute for expansion in breadth—indeed, they may create problems on the breadth dimension, as when artificial intelligence eliminates existing jobs and widens income disparities. This explains a common puzzle: why growth remains under heavy pressure even though innovation-driven development and new industries appear to be doing well.

Here it is necessary to introduce the concept of end demand. End demand refers to the portion of GDP that does not re-enter the next production process. It comprises all consumption plus non-productive—or consumption-oriented—investment, chiefly investment related to people’s livelihoods in real estate, infrastructure, and other services. Viewed from the natural flow of all economic activity, end demand constitutes final products in the true sense: the purposive portion of GDP devoted directly to meeting people’s aspirations for a better life. GDP minus end demand equals productive investment plus net exports—both of which are instrumental, existing to serve the growth of end demand.

For a long time, China’s consumption share of GDP has been roughly 20 percentage points below the international average—a structural deviation. But because real estate and infrastructure grew rapidly for an extended period, this deviation was effectively masked. After 2022, as real estate slumped sharply and infrastructure investment decelerated, the long-hidden structural shortfall in consumption was laid bare, becoming the choke point within end demand.

In recent years, the growth of end demand has slowed markedly and contracted in relative terms, dragging down the GDP deflator and pushing nominal growth below real growth. Combined with excessive investment—especially productive investment—this has driven capacity utilization down. When capacity utilization is too low, returns on investment are correspondingly poor, and investors become unwilling to invest more. This is the main reason for the current slide in investment.

Growth in industrial value added has held up reasonably well, driven mainly by exports and, more recently, by rising capital expenditure on artificial intelligence. But in the system of national accounts, net exports count as savings and correspondingly reduce domestic consumption. Strong exports are thus one cause of weak domestic consumption—a linkage that has not received the attention it deserves.

III. Making the Savings–Consumption Formation Mechanism the Focus and Point of Breakthrough

Behind “strong supply, weak demand” lies a national income structure of high savings and low consumption. By global standards, China’s savings rate is conspicuously high. Over the past decade or so it peaked above 50%, and though it has edged down in recent years, it remains above 42%. By contrast, the global average savings rate is roughly 25%. In terms of composition, China’s savings come mainly from the corporate and household sectors; government savings have gradually declined and have turned negative in recent years.

Take 2022 as an example. China’s flow-of-funds accounts show that GDP that year was roughly RMB 120 trillion, with total savings of RMB 55 trillion—a savings rate of 46% and a corresponding consumption rate of about 54%. Corporate savings amounted to RMB 27 trillion and household savings to RMB 27.6 trillion, roughly half each.

Corporate savings reached 22.5% of GDP—markedly high by international comparison. This is directly related to the low share of dividends in corporate capital returns. In 2022, only 10.2% of Chinese households’ property income came from corporate dividends, well below the global average of 55.7%. Historically, during the high-growth era, investment funding needs were enormous while external financing was scarce and costly, so firms leaned toward self-financing and retained more earnings. More importantly, state capital accounts for a large share of total corporate capital, and state-owned enterprises long paid little or nothing in dividends. The state capital operating budget system introduced in 2007 changed this somewhat. As SOE reform has deepened and the governance of state capital has improved, dividend payouts by state capital in listed companies have increased, but a large number of unlisted SOEs still pay little. With corporate income taking an outsized share of society’s total disposable income and being converted into corporate savings, the economy-wide corporate savings rate has ended up high. And corporate savings have only one use: investment.

Household savings are also elevated, but looking inside the household sector—ranking households by income from high to low—one finds that it is chiefly the massive savings of high-income groups that push up the overall household savings rate. China Merchants Bank’s 2024 annual report shows that its “Sunflower” clients (those with average daily assets of RMB 500,000 or more) accounted for just 2.49% of its retail customers yet held 81.90% of assets; compared with 2023, new assets from Sunflower clients made up 87.48% of all newly added assets. This indicates that a small number of high-net-worth clients hold the overwhelming majority of the bank’s savings assets, with concentration still rising. Data from other banks show similar patterns. Some argue that since China adds well over RMB 10 trillion in new savings each year, it would suffice simply to guide households to convert these savings into consumption. This view—focused on aggregates while ignoring structure, and blind to the direct link between high savings, low consumption, and excessive income disparities within the household sector—is deeply misleading both in theory and for policy.

IV. How to Advance Reform of the Savings–Consumption Formation Mechanism and Adjust Policy

Investment has now been contracting for more than a year, and in May this was compounded by negative growth in retail sales of consumer goods—something unseen since reform and opening up began, and which should command the highest attention. China’s economy once again stands at a critical turning point. The urgent task is to restore investment vitality and return investment to positive growth. The way to do so is to substantially lower the savings rate and correspondingly increase end demand. This differs from what some believe, and it is the key to understanding how China’s economy operates at the current stage.

National income divides into savings and consumption. If high savings do not come down, low consumption cannot go up—a simple, hard constraint. Only by lowering the savings rate and increasing end demand can capacity utilization be raised; and once utilization reaches a relatively high level, genuine demand for investment will emerge—precisely the “investment backed by demand and returns” that the central authorities have repeatedly emphasized. If instead the savings rate is not lowered, end demand is not increased (or is in effect reduced), and investment is forced up by the old methods, the supply–demand imbalance will only intensify, steering China’s economy in an altogether different direction.

The point of breakthrough for correcting the imbalance lies in reforming the savings–consumption formation mechanism and converting excess savings into effective consumption. Several priority areas of reform, with corresponding quantitative policy targets, can be proposed.

First, within a fairly short period—say three years—bring the economy-wide savings rate below 40% and the corporate savings rate below 20%, converting part of corporate savings into consumption and raising the economy-wide consumption rate above 60%. This would also be a concrete step toward implementing the 15th Five-Year Plan’s goal of raising the household consumption rate. The direct route is to raise corporate dividend payout ratios; transferring a sizable share of state capital to the social security fund is, in effect, one concrete form of state-capital dividend distribution.

Some harbor doubts about how state capital could be transferred: not much state capital is genuinely profitable and able to pay dividends; in some sectors state capital carries heavy debt burdens, making net assets hard to pin down; state capital falls under different departments and agencies, so transfers to the social security fund would meet resistance; and so on.

These problems are in fact not hard to solve. Simply put, the transfers can be made from state capital that is already listed. The A-share market is currently worth roughly RMB 120 trillion, of which state capital is preliminarily estimated at around 40%. A certain amount—say RMB 20 trillion—could be transferred from this into the social security fund. The transferred state capital would not change its departmental or institutional affiliation; instead, new pension funds would be established under the existing departments and agencies to manage the transferred capital, with new operating functions and objectives. From the capital market’s perspective, this merely converts a portion of state capital into pension assets, adding more long-term, patient capital.

Private enterprises should also be encouraged to raise dividend payouts. Through improved corporate governance, stronger protection of minority investors’ rights, and the promotion of long-term value investing, firms should be encouraged to return more profits to shareholders as dividends, which shareholders can then convert into consumption spending.

Second, steadily advance tax reform to curb the rise of the Gini coefficient and gradually bring it down to 0.4 or below, thereby raising society’s average propensity to consume and lowering the household savings rate.

International experience shows that economies that successfully moved from middle-income to high-income status mostly had Gini coefficients between 0.3 and 0.4, some even lower. A more balanced income distribution channels more national income into end-demand spending, easing the pressure of releasing the large capacity built up during the high-growth era.

China’s current tax structure is dominated by indirect taxes, with direct taxes such as personal income tax and property taxes accounting for a low share. From the standpoint of reforming the savings–consumption mechanism, reasonable taxation of high-income groups should be increased—focusing on personal income and property taxes—thereby raising government revenue. Some worry that this would damage the confidence and expectations of high earners. In reality, the negative effects would be limited; if anything, it would strengthen the government’s incentive to protect property rights. As one formulation puts it, the boundary of the capacity to tax determines the boundary of property-rights protection—because protecting property rights means protecting the tax base.

Third, shift the center of gravity of macroeconomic management and government functions at all levels from the supply side to the consumption side. More than 50% of macro stimulus spending should be devoted to addressing the supply–demand imbalance, with the emphasis on building the social security system for low- and middle-income groups.

There is some public confusion about how consumption drives growth. Reforming the savings–consumption mechanism and raising the incomes and consumption of low- and middle-income groups would, first, create new consumer market space, turning excess capacity into adequate—perhaps even insufficient—capacity, and thereby spurring new, effective investment. Second, and more importantly, expanding consumption enables more people, especially low- and middle-income groups, to share in the fruits of reform, opening up, and development, deepening their sense of gain as their aspirations for a better life are met. Equally important, expanding this group’s consumption—above all developmental consumption—is in essence investment in human capital, a concrete embodiment of “investing in people,” providing the human-capital foundation most vital to innovation-driven growth.

In raising the incomes and consumption of low- and middle-income groups, subsidies for goods purchases can accomplish only so much. The focus should be on two glaring shortfalls: housing and social security.

A migrant worker may own a 200-square-meter two-story house in the countryside, yet return there only a few days a year, spending the rest of the time crammed into a 10-square-meter basement room in the city. Statistically, his available housing space totals 210 square meters, but his actual housing conditions are dire. Migrant workers and similar groups face an enormous structural mismatch in housing between urban and rural areas. Housing-security policies and the reallocation of urban–rural resources should be used to accelerate solutions to the housing difficulties of migrant workers and other new urban residents—which would also supply genuine demand to help the real estate sector stabilize and return to normal growth.

Through the transfer of state capital to social security, fiscal subsidies, short-term stimulus funding, and improvements to social insurance contributions, the aim during the 15th Five-Year Plan period should be a marked increase in per capita pension income under the basic pension scheme for urban and rural residents—the scheme covering the largest number of low- and middle-income people—commensurate with the current imperative of boosting end demand and stabilizing macroeconomic growth. Among the many consumption-stimulus measures now on the table, this reform ranks among the most effective for driving growth and plays an irreplaceable role in stabilizing it.

Some contend that raising pension incomes for rural elderly people would do little to expand consumption. This is a specious argument. In fact, the excessively low pension incomes of those covered by the urban–rural residents’ pension scheme suppress society’s consumption potential in multiple ways. First, they directly constrain the spending power of 170 million pension recipients. Second, inadequate pensions for rural parents force their children to provide larger intergenerational transfers, crowding out the children’s own consumption. Third, they weaken the future expectations of the 370 million people currently contributing, pushing them into higher precautionary savings. The urban–rural residents’ pension scheme covers 550 million people—95% of them rural residents—accounting for roughly half of all pension participants nationwide. It is the segment of China’s pension system with the lowest incomes, the largest scale, and the highest propensity to consume. Seen in this light, the pension shortfall for this population constrains the consumption growth of roughly half of China’s people, with economy-wide implications that cannot be ignored.

V. Clarifying Some Misconceptions About Reversing the Imbalance

One view holds that the production-first, consumption-second growth model has existed for a long time—we have long spoken of it as unbalanced, uncoordinated, and unsustainable, yet we got through all those years, so perhaps it can continue. But in the past we were long in a shortage economy, and even after entering the demand-constrained stage there was still room for investment. Those conditions no longer exist. Investment has been contracting for an extended period—evidence that sustaining this growth model is becoming ever more difficult.

Another view holds that expanding consumption and raising the incomes of low- and middle-income groups is a livelihood issue that should proceed within our means and without haste. For a long time this view had some merit. But the situation today is that without raising these groups’ incomes and expanding end demand, capacity utilization cannot improve, investment cannot resume growth, and actual growth will fall well below potential. Raising the incomes and consumption of low- and middle-income groups is a livelihood issue, yes—but even more, it is an urgent matter of macroeconomic stabilization. In the past, when growth flagged, we reached for investment; at the current stage, we must reach for consumption. The logic is the same; only the times have changed, and with them the areas holding growth potential.

Still another view holds that while structural and institutional reforms matter, they are distant water that cannot quench present thirst, so the short run must rely on macro stimulus. In fact, structural reform and macro policy each have their functions; they are not an either/or choice, and combining them often works better. Some structural reforms—such as the transfer of state capital to social security described above—can deliver results in the short term. Others, such as reducing the Gini coefficient, may look slow in the short run once targets are set and action begins, but prove fast in the long run. Substituting short-term stimulus for structural reforms that could actually be advanced—or using it to avoid them—will only prolong the problems and raise their cost. Japan was the first country to propose and implement quantitative easing and zero or negative interest rates. More than twenty years on, it offers no record of decisively escaping deflation or restoring growth vitality. Many in recent years have studied Japan’s experience; China should draw lessons from it rather than repeat its mistakes.

On reversing the imbalance between strong supply and weak demand, the central leadership’s judgment is clear, and the path to a solution is equally clear. International experience shows that economies caught in the “middle-income trap” mostly suffered from weak innovation and uncompetitive industries; China’s shortcoming today, by contrast, is “strong supply, weak demand.” Logically, demand-side problems are relatively easier to solve than supply-side ones—but they are not without difficulty. The imbalance is deeply bound up with a web of institutional and structural factors. Breaking out of the old growth framework—shifting from growth driven mainly by investment and exports to growth driven mainly by innovation and consumption, and swiftly establishing a new growth framework—will require a new round of intellectual emancipation and a new social consensus. It means casting off path dependence and confronting and resolving the necessary transformations in thinking, interest relationships, policy instruments, and working methods, so that strong supply propels strong demand, a new cycle of dynamic mutual reinforcement between supply and demand takes shape, the macroeconomic fundamentals are stabilized, and sustained momentum is secured for long-term, stable growth.

More to read

The East is Read

Liu Shijin Makes a Forceful Case for Raising Rural Pensions

Liu Shijin is a former Vice President (Vice Minister) of the Development Research Centre (DRC), a comprehensive policy research and consulting institution directly under the State Council, the central government of the People’s Republic of China. In recent years, he has been one of the most senior voices urging Beijing to raise the meagre pensions paid …

Read more

2 months ago · 7 likes · 1 comment · Zhu Yutao and Yuxuan JIA

China's position can be summarized as “non-acceptance, non-participation, non-recognition, and non-implementation.” The rationale is that the dispute is essentially about territorial sovereignty, which falls outside the scope of the UN Convention on the Law of the Sea (UNCLOS) and therefore beyond the tribunal’s jurisdiction.

For today’s episode, I want to present the latest remarks from Professor Wu Shicun (吴士存) at the South China Sea Security Roundtable Dialogue. Wu is the founding president of the National Institute for South China Sea Studies. From 1993 to 2000, he took part in the China–Vietnam negotiations on the delimitation of the Beibu Gulf (Gulf of Tonkin) as a member of the Chinese government delegation and its expert group. He has also studied or pursued advanced training at the Johns Hopkins School of Advanced International Studies (SAIS), the Asia-Pacific Center for Security Studies, and Harvard University's Kennedy School of Government. He is one of China's most authoritative experts on the South China Sea issue.

In his latest commentary, Wu argues that the award lacks legal validity, has been increasingly instrumentalized for geopolitical ends, and constitutes an obstacle rather than a pathway to peace and stability in the South China Sea. As Manila marks the anniversary with renewed diplomatic and academic activity around the ruling, the article contends that the next decade will see continued contestation over its legacy and meaning.

Thanks to Professor Wu's kind authorization, I can present the English version in my newsletter.

The “South China Sea Arbitration Ruling” Is a “Bad Precedent” That Violates International Law

Ten years after the illegal ruling in the “South China Sea arbitration case,” unilaterally initiated by the Philippines, the task of refuting and detoxifying it remains arduous.

A few days ago, Manila hosted a series of events related to the illegal ruling. On the surface, these were held in the name of think tanks or academic conferences, but in substance, they were led by the Philippine Department of National Defense, Department of Foreign Affairs, or national security agencies. Looking at the nationalities of the participants, virtually none were absent among the countries supposedly “like-minded” with the Philippines, countries with maritime disputes with China, and countries seeking to exploit the South China Sea for geopolitical gain. At these meetings, certain Philippine politicians and “scholars” with vested interests tied to the Philippines proclaimed that their goal for the next decade on the “ruling” would be to translate the so-called “legal achievement” into actual strategic advantage through coordination among think tanks, the Philippine government, the military, and civil society. Philippine Defense Secretary Teodoro, who has been sanctioned by China, went so far as to declare nakedly that legal claims must be closely integrated with military buildup to serve an active defense posture and a narrative of national unity.

Clearly, this illegal and invalid ruling has been further distorted. It has become the bond and glue for interest-binding between the Philippines and certain countries seeking to manufacture a “China threat in the South China Sea” narrative. Japan and a few other countries that, like the Philippines, wish to package and “launder” illegal gains through illegal means are also taking part. It is fair to say that the countries trying to get a share of the illegal ruling have formed a de facto “interest group” and an interlocking “industrial chain.”

Precisely for this reason, while intensifying our legal refutation and exposure of the ruling’s fallacies, we must also take a multi-pronged approach: sever the interest chain built on this illegal ruling and reveal to the world the underlying logic of the “interest group” formed around it.

Looking back at the ten years since the illegal ruling was issued, the Philippines and the United States and other Western countries have sung in chorus, fabricating absurd logic and false narratives such as “the ruling is international law” and “failure to implement the ruling is a violation of international law,” and repeatedly hyping them for fear that the “ruling” would be forgotten or even “buried” by the international community. For ten years, this illegal ruling has been like a terminally ill patient, barely kept alive by life-support systems in the ICU. Judging from the Philippines’ current frequent provocations against China’s rights and claims in the South China Sea and its actions undermining regional peace and stability, the current Philippine government is intensifying its use of the “ruling” to change the status quo, entrench illegal gains, and attempt to expand the scope of its infringements. Evidently, the illegal ruling has had an extremely negative impact on political mutual trust and overall relations between China and the Philippines. Beyond the Philippines, a few other parties to South China Sea disputes have also occasionally cited or invoked the “ruling” as a basis for their claims and unilateral encroachments. Some countries outside the region have lavished praise on the illegal ruling, attempting to use it to pressure China into “complying with” or “implementing” it.

The fact that the illegal ruling has caused instability and localized turbulence in the South China Sea shows that it has not brought — and can never bring — peace and tranquility to the region. On the contrary, it has become a “troublemaker” and a “source of turmoil” for peace and stability in the South China Sea, a “spoiler” of bilateral relations between China and other claimants on South China Sea issues, and the greatest obstacle and “stumbling block” to the full and effective implementation of the Declaration on the Conduct of Parties in the South China Sea and the orderly advancement of consultations on a Code of Conduct.

From the above analysis, we can draw a conclusion: unless the illegal ruling is laid to rest, there will never be peace in the South China Sea. Unless we say a firm and thorough “no” to the illegal ruling, maritime cooperation in the South China Sea will remain merely verbal and on paper. Unless the illegal ruling disappears from public discourse on the South China Sea, China will from time to time be labeled by ill-intentioned countries or forces as “undermining the rules-based international order.” And unless the illegal ruling is uprooted entirely, other countries may follow suit, using operations that violate international law as cover and sophistry when committing infringements, provocations, and harm to public interests in the South China Sea.

At the same time, it remains essential to expose, at the root, the illegal nature of the “South China Sea arbitration tribunal” on the basis of international law, international practice, and historical facts, and to refute the political conspiracy behind it and its absurd claims of “jurisdiction.” Otherwise, some countries in the international community that do not know the truth, or observers of the South China Sea issue, may be misled by false narratives such as “the ruling is international law” or “the ruling is part of international law,” which over time could solidify into part of their perception of the South China Sea.

Looking back over the past decade, one heartening reality is that although the illegal ruling attracted some international attention when it was first issued in July 2016, its content has received no response at the level of state practice from other countries, no recognition or adherence from international judicial bodies, and no endorsement from international academia. Some countries that support the illegal ruling know in their hearts that it is illegal and invalid; institutions such as the International Court of Justice have collectively avoided substantive discussion of the “ruling” in relevant cases; and upright and conscientious authoritative international scholars have leveled sharp criticism at the “ruling” from multiple angles, including textual interpretation, historical examination, and logical consistency.

In short, the illegal ruling in the “South China Sea arbitration case” will not produce any substantive impact at the level of international law of the sea; it is primarily a tool used by certain countries or forces to stir up trouble in the South China Sea. In the foreseeable future, the “ruling” will increasingly be seen as an “isolated incident” within a specific geopolitical context, rather than a so-called “precedent” in the international law of the sea system. In any case, over the next ten years or longer, we will not relax our vigilance against any claims or actions based on this illegal ruling, and we will sternly refute, lawfully defend our rights against, and resolutely counter any words or deeds that hype it.

As a Beijing resident, the story hits close to home. The Beijing I remember was dry and hot. But since moving back in 2021, I've watched the city grow noticeably wetter. No year from my childhood was ever as rainy as the past two have been. And it’s not just my feeling. Last month, Caixin reported that Beijing’s average annual precipitation over the past decade has risen by twenty percent, with rainfall in 2025 exceeding 900 millimeters.

For anyone educated in China, that number calls to mind a concept from high school geography that Beijing sits right on the 400-millimeter isohyet—the line connecting places that receive 400 millimeters of annual rainfall. This line is about far more than weather. It marks the boundary between two ways of life that defined traditional China, between farming and nomadic civilizations. Beyond it, rainfall is too scarce to sustain agriculture, and people historically survived by herding instead. For that reason, successive Chinese dynasties treated the line as a natural defensive frontier.

Now, that line appears to be shifting. As a history enthusiast, I couldn't help wondering what this might mean, and I’m not the only one. Professor Wang Zhengxu of Zhejiang University has been thinking along the same lines. The other day, he shared with me an essay he wrote in 2023 about how rainfall shaped the social, political, and economic patterns of ancient China. I found it illuminating, and I'd like to share it with my readers here.

Subscribe now

Professor Wang Zhengxu (王正绪) is a distinguished professor at the School of Public Affairs, Zhejiang University. Previously, he served as a distinguished professor at Fudan University, senior fellow and acting director at the China Policy Institute, University of Nottingham, as well as a research fellow at the East Asian Institute of the National University of Singapore. His research spans comparative politics, Chinese politics, ideas and concepts of political science and East Asian state-making.

It's a long article with many Chinese names and historical events. Basically, it discusses Professor Wang's answer to the "Great Divergence" problem, which asks why capitalism and the Industrial Revolution emerged in Europe rather than in China. The traditional answer is that Europe's fragmented system of rival states drove technological, military, and fiscal innovation, while a unified China after the Qin and Han lacked that competitive pressure. But Professor Wang pushes back on this theory. East Asia, he argues, was hardly short of brutal interstate competition. The Wei-Jin era produced a sixteen-state system in North China; even the mighty Tang saw its capital fall to rebellion and Tibetan assault; and during the Song dynasties, the rivalry among Song, Western Xia, Liao, Jin, and the Mongols routinely ended in the annihilation of entire states. Competition was not the missing variable.

What was different, Wang contends, was the character of the state itself. The Confucian-Legalist state (儒法国家) looks inward rather than outward, seeking to sustain itself rather than plunder, which is why China produced Zheng He’s voyages but no Columbus's state-sponsored conquest of distant continents. And the sheer mass of China’s agrarian economy gave it a resilience that made conquest ultimately self-defeating: dynasties could retreat south of the Yangtze and flourish, while nomadic conquerors who crossed the Wall were absorbed into the agrarian order they had invaded. Measured against this civilizational bulk, Wang concludes, East Asia’s interstate wars remained storms in a teapot.

Thanks to Professor Wang's kind authorization, I can share this article in English:

Below is the full article:

Geography, Climate, and Long-Term Patterns of Socioeconomic History: The Hu Huanyong Line in History and on the Ground
地理气候与长期的社会经济历史模式——胡焕庸线的历史与现场

Natural factors such as geography and climate are structural conditions for the development of human societies. Human communities form their economic, political, and cultural patterns of development in accordance with the natural geographic conditions on which they depend. In the long sweep of Chinese history, geography and climate produced a famous economic, social, and cultural dividing line: the Hu Huanyong Line (胡焕庸线). Running from China’s northeast to its southwest, this line broadly separates the pastoral-nomadic regions of East Asia from its agricultural regions. In terms of physical geography, this boundary between herding and farming roughly coincides with the 400mm isohyet (400毫米等降雨线) — the line of equal annual rainfall — on the East Asian continent. The course of the Great Wall (长城), whose construction began in the pre-Qin era, largely follows this isohyet, making it the “institutional” boundary between the agricultural and pastoral zones of the ancient Chinese world. The Hu Huanyong Line, or the 400mm isohyet, is therefore an essential frame of reference for understanding the long-term history of East Asia.

This article addresses three questions. First, in what ways did this boundary between human economic and social modes — a boundary created by natural geography — shape the long-term patterns of economic and social development in East Asia? Second, does this natural-geographic factor continue to influence and shape economic and social development in the industrial age? Third, does the climate change now unfolding across the globe require us to re-examine the influence of this geographic dividing line on East Asian society? As its object of study, this article selects the region along the Great Wall between Pingxing Pass (平型关) and Yanmen Pass (雁门关) in northern Shanxi Province (山西省), attempting to answer the questions above through an analysis of this particular locality. The region straddles both sides of the Great Wall in North China, with the Hu Huanyong Line — the 400mm isohyet — passing directly through it. Historically, it was a crucial meeting point between the northern pastoral world and the agrarian civilization of the Central Plains (中原). In antiquity it fell under the administrative units of Daizhou (代州) and Yanmen Commandery (雁门郡), as well as Shuozhou (朔州) and Yunzhou (云州) north of the Wall; this article therefore calls it the “Yan-Yun Zone” (雁云地带). The zone principally comprises the city of Xinzhou (忻州) south of Yanmen Pass and the cities of Shuozhou and Datong (大同) to its north. Historically, the nomadic polities of the north and the agrarian polities of the south experienced military conflict here, but also exchanges of culture and ways of life, along with flourishing commerce. All of these forces shaped the highly distinctive economic and social patterns of northern Shanxi on both sides of the Great Wall. Combining historical texts, findings from general social science theory, and on-site observation, this article uses methods akin to thick description and field notes to paint a holistic portrait of the Yan-Yun Zone. It offers a detailed on-the-ground description and verification of the theoretical proposition that “three lines converge as one” — the Great Wall, the Hu Huanyong Line, and the 400mm isohyet — and, drawing on multiple disciplines, proposes a new interpretive framework. It seeks to demonstrate the important role of the Hu Huanyong Line in the evolution of the civilizational order of the East Asian continent, and its long-term shaping of historical socioeconomic patterns of agro-pastoral contact, military confrontation, and frontier trade, thereby moving beyond the single-discipline perspectives of demography or economic geography that have dominated previous research on the line. The article also examines the potential role of the 400mm isohyet in future population migration and socioeconomic development against the backdrop of global climate change, in the hope of broadening the scholarly dimensions and theoretical substance of geopolitical research, embodying the dual values of theoretical innovation and engagement with real-world concerns.

I. Natural-Geographic Social Units: Questions and Methods

In Aristotle’s view, the natural environment and its climatic features nurture heterogeneity in civic character; geography also shapes citizens’ preferences and the formation of polities. Montesquieu regarded human life as a reflection of geographic and climatic conditions; Herder analyzed the distinctiveness of different national cultures on the basis of geographic factors, famously asserting that “history is geography in motion.” Jean Bodin’s Six Books of the Commonwealth likewise contains a dedicated exposition of climate theory. The British geographer Halford Mackinder held that the geographic causation of politics is “permanent, inevitable, and omnipresent.” The east-west orientation of the Eurasian landmass allowed pathogens, crops, and domesticated livestock to diffuse across vast distances, profoundly shaping the forms taken by human civilization. By contrast, the north-south orientation of the Americas constrained the spread of crops and fettered the development of their ancient civilizations. The state of Qin (秦国) unified the territory of ancient China between the Yangtze River (长江) and the Yellow River (黄河), whereas Napoleon, for all his military might, could not unify Europe; the geographic differences between East Asia and Europe were a decisive factor. Ray Huang (黄仁宇) emphasized the decisive role of geographic factors — soil, wind patterns, rainfall — in the political development of ancient China, and he perceptively noted the influence of the 400mm isohyet on the civilizational system of the East Asian continent. Zhao Dingxin (赵鼎新), in his research on state formation in pre-modern China and on the interactions between nomadic and agrarian peoples, likewise assigns a central place to geographic-environmental factors.

The Hu Huanyong Line is a paradigmatic example of how geography shaped human social development on the East Asian continent. In 1935, Hu Huanyong (胡焕庸), using one dot to represent twenty thousand residents, hand-plotted more than twenty thousand dots on a map of China and drew isopleths to produce a population-density map of the period. The line connecting the two points of Aihui (瑷珲) and Tengchong (腾冲) on that map came to be known as the Hu Huanyong Line. To the northwest of this remarkable line lies the land of “white horses in the autumn wind” — a relatively arid region of grassland and desert dominated by animal husbandry, which at the time held only 4% of the national population. To the southeast lies the land of “apricot blossoms in spring rain” — a region of relatively abundant rainfall and developed agrarian civilization, which then contained 96% of the country’s population. By the Seventh National Population Census of 2020, China’s level of economic development, industrialization, urbanization, and population structure and distribution had been utterly transformed compared with the 1930s, when the line was first proposed — yet the population ratio between the northwestern and southeastern halves demarcated by the Hu Huanyong Line still stood at 6.5% to 93.5%. Strikingly, Ray Huang also regarded the 400mm isohyet as a major dividing line across the Chinese land: to its southeast lies a monsoon climate, with settled agriculture as the principal mode of production and social organization; to its northwest lies a temperate continental climate, with animal husbandry as the dominant mode. Huang further observed that the course of the Great Wall overlaps with the 400mm isohyet over a very wide extent. Although the isohyet shifts somewhat from year to year, it and the Great Wall have throughout history marked the boundary between the agrarian civilization of the Central Plains and the nomadic civilization of the north. Moreover, the 400mm isohyet also broadly coincides with the Hu Huanyong Line.

An examination of recent rainfall maps shows that the 400mm isohyet essentially coincides with the stretch of the Great Wall between Pingxing Pass and Yanmen Pass. Traveling north from Xinzhou and passing just beyond Yanmen Pass, one crosses the 400mm isohyet. North of this point lies the historical Yunzhou, precisely the region where the Northern Wei dynasty (北魏) arose. This dynasty, founded by the Tuoba Xianbei (拓跋鲜卑), moved its capital from Datong to Luoyang (洛阳) under Emperor Xiaowen (孝文帝), writing a chapter of history in which a steppe people voluntarily sinicized and transformed itself into a Chinese dynasty. The Yan-Yun Zone of northern Shanxi, lying along the Hu Huanyong Line and the 400mm isohyet, is precisely where the arid and humid zones, frontier and pastoral-agrarian ecologies, and nomadic and agricultural civilizations of the East Asian continent converge. Throughout history it served as a corridor and channel for the exchange and collision of Central Plains civilization with the nomadic civilization of the northern steppe. From King Mu of Zhou’s (周穆王) frontier tours to Li Mu (李牧) of the state of Zhao (赵国) crushing one hundred thousand Xiongnu (匈奴); from the campaigns of Meng Tian (蒙恬), Wei Qing (卫青), and Huo Qubing (霍去病) against the Xiongnu to Wang Zhaojun’s departure beyond the frontier (昭君出塞); from Su Wu herding sheep in captivity (苏武牧羊) to Cai Wenji’s return to the Han (文姬归汉); from King Wuling of Zhao (赵武灵王) adopting nomad dress and mounted archery (胡服骑射) to Emperor Xiaowen’s relocation of the capital, and on to the Yang Family Generals (杨家将) resisting the Liao (辽) armies — all of these underscore the region’s character as a meeting point of civilizations.

This article is based on first-hand materials — field observations, interviews, and records — gathered by the author and a study group during a visit to the Yan-Yun Zone in May 2023. It offers a relatively comprehensive descriptive account of the research object and examines existing theoretical narratives of geographically defined social units, aiming at a descriptive interpretation of this complex cultural setting. In the process of narration and description (that is, representation), the subjective impressions, reflections, and puzzlements of the research team members on site are appropriately combined with records and representations of objective reality, including exchanges and interactions between researchers and local colleagues or contemporaries at different sites. Such “reflexivity” acknowledges the connection and mutual influence between observer (researcher) and object of study; it acknowledges the interactive relationship between social reality — what, for instance, the Great Wall actually is, or what historical meanings Yanmen Pass carries — and the researcher, recorder, and narrator, which inevitably means that different researchers will produce different research records. Our itinerary covered the cities of Taiyuan (太原), Xinzhou, and Datong, and included sites that display the principal historical and cultural features of northern Shanxi in each historical period: Pingxing Pass, Yanmen Pass, the Yungang Grottoes (云冈石窟), the old city of Datong (大同古城), the Ming-era Great Wall at Yanggao (阳高) and the Shoukou Fort (守口堡), and the Hanging Temple (悬空寺), along with former residences of famous figures and memorials of the War of Resistance and the revolution. We surveyed the region’s major rivers, lakes, reservoirs, mountains, plains, and basins, and at one point crossed beyond the Yan-Yun Zone into the Ulanqab (乌兰察布) region of Inner Mongolia (内蒙古). We held long, unstructured conversations and discussions with colleagues who have long lived and worked locally — university teachers, township leaders, retired cadres of state-owned enterprises, museum staff, compilers of county gazetteers, and officials of propaganda departments — who provided us with rich information and diverse perspectives.

II. Rainfall and Its Uneven Distribution: The Divide between Farming and Herding

Central and northern Shanxi rest upon the deep, vast loess lands that nurtured several millennia of Chinese civilization. Traveling north from Taiyuan in central Shanxi, one gradually senses the diminishing surface water and intensifying aridity. For travelers from the south, the most immediate experience of the low-humidity climate is often waking to find their nasal passages so dry they bleed; clothes washed before bed are completely dry by morning. When local colleagues learned that our study group came mainly from the Jiangsu-Zhejiang-Shanghai region (江浙沪), their first thought was of their own discomfort in the humid climates of Hangzhou (杭州) and Shanghai (上海), perpetually drenched in sweat — “you have to shower the moment you step indoors.” A dry climate may be pleasant, but Shanxi, Shaanxi (陕西), and indeed all of North China are constantly threatened by water scarcity and drought — a far more consequential manifestation of climate’s influence on human society. Numerous prefectural and county gazetteers of Shanxi repeatedly record the severe blows dealt by drought to local agriculture and social life, and local colleagues all remembered summers in the past when the urban water supply was cut off.

Agriculture places demands on climate that are, in a sense, quite exacting, and precipitation is the key constraint. It is generally held that, given suitable sunlight, soil, and terrain, an annual precipitation of 250 to 400 millimeters provides the minimum basis for crop cultivation. China lies in a monsoon zone, and rainfall diminishes gradually from southeast to northwest. The many mountain ranges between the coast and the interior form natural barriers to the northwestward movement of warm, moist air. Thus a line running along the Greater Khingan Range (大兴安岭), through the Yin Mountains (阴山), the Helan Mountains (贺兰山), and the Bayan Har Mountains (巴颜喀拉山), and extending southwest to the Gangdise Range (冈底斯山), forms the 400mm isohyet — the divide between the semi-humid and semi-arid zones of the East Asian continent.

Traditionally, this isohyet marked the boundary between crop agriculture and steppe-nomadic civilization on the East Asian continent. Northwest of the line, constrained by precipitation and temperature, the institution of animal husbandry had to be interposed between humans and the grassland in order to sustain human social organization of any scale. Cattle, horses, and sheep were the steppe tribes’ principal source of food and also of clothing, footwear, tents, and other necessities. Horses, moreover, enabled steppe armies to conduct long-range raids and distant campaigns, giving them an advantage in attacking and plundering the settled agricultural units south of the Great Wall and on both sides of the Tianshan Mountains (天山). Because food could come only from livestock, the steppe’s population-carrying capacity was merely 1–2% of that of agricultural regions.

The nomadic way of life, moreover, is acutely sensitive to climatic change. Even a slight drop in average global temperature drastically shortens the frost-free season in pastoral regions and significantly raises the probability of blizzards and other disasters, causing livestock to starve or freeze in great numbers. Cooling climates thus gave nomadic peoples enormous impetus to migrate southward. Historical research shows that whether it was the Xianyu (鲜虞), Rong (戎), and Di (狄) entering the Central Plains in the pre-Qin era, the northwestern tribes and northern Xianbei (鲜卑) migrating inward during the Wei-Jin period (魏晋时期), or the rise, expansion, and eventual conquest of the Central Plains by the Western Xia (西夏), Jin (金), Liao, Yuan (元), and Qing (清) dynasties, all occurred during cold periods in Chinese history.

The Yan-Yun Zone sits at the edge of the 400mm isohyet, the northern limit of the semi-humid region. Here one can feel how deeply human survival is constrained by natural conditions. Given the limited surface water, the principal grain crop north of the Huai River (淮河) in China is wheat — yet wheat was first domesticated in the Fertile Crescent of Western Asia before spreading to Europe and across Asia, whereas Shanxi is the homeland of millet. In the Yellow River valley, the earliest grain cultivated by the ancestors of the Chinese was very likely millet. Millet was anciently called ji (稷), and the emergence of the term sheji (社稷, “the altars of soil and grain”) marked the appearance of human organization in the form of the state. Traveling through Shanxi, one finds local people speaking with relish of the province’s agricultural specialty: minor coarse grains (小杂粮). Reportedly, Shanxi accounts for roughly one-tenth of China’s coarse-grain acreage, and in recent years Xinzhou has produced one-third of Shanxi’s output — hence the saying, “China’s coarse grains are in Shanxi; Shanxi’s coarse grains are in Xinzhou.” As China’s food security has improved, coarse grains such as naked oats, buckwheat, quinoa, foxtail millet, proso millet, and maize, along with mung beans and adzuki beans, have taken up an ever larger share of the Chinese staple diet. Our study group visited many counties, cities, and townships, and wherever we went, every proper meal featured a large plate of golden fried cakes made of proso-millet flour (黄米面油炸糕) — an indispensable local dish for honoring guests.

Climate furnished two structural conditions for the region’s historical development. First, limited rainfall severely constrained the land’s population-carrying capacity and the robustness of its ecosystem. Field observation shows that in recent years numerous river-management, surface-water protection, and ecological construction projects have begun to bear fruit; the flows of major rivers such as the Fen River (汾河) and the Hutuo River (滹沱河) are now well protected. Like many other Chinese cities, Taiyuan, Xinzhou, and Fanshi (繁峙) now have places for boating and fishing and riverside wetlands, and the living environment has improved enormously. Historically, however, years of severe drought could bring crop failure and famine, and ecological pressure produced the tragic, heroic memory of “Going West of the Pass” (走西口) in northern Shanxi. The “West Pass” (西口) was the famous frontier town of Shahukou (杀虎口) in Youyu County (右玉县), Shuozhou; “going west of the pass” meant leaving through it to seek a livelihood north of the Great Wall. Second, the region’s position at the edge of the Hu Huanyong Line also meant it lay at the junction of the agrarian and nomadic worlds. Such a geopolitical and geo-economic location meant that the Yan-Yun Zone lived for long periods under threat from the Xiongnu, the Mongols, and other nomadic peoples — yet in peacetime it enjoyed superb conditions for frontier trade.

To this day we can see the traces left by these two structural conditions. The region preserves monuments and relics from the eras of nomadic regimes — the Northern Dynasties (北朝), Liao, Jin, and Yuan — and the murals of the Northern Dynasties, Jin, and Yuan displayed in local museums are breathtaking. During our fieldwork we repeatedly “encountered” the former residences, footprints, deeds, and works of the two literary giants Yuan Haowen (元好问) and Bai Pu (白朴), evoking the literary splendor of the Jin and Yuan eras. Local people are deeply familiar with the lives and works of both men; many can recite offhand Yuan Haowen’s immortal line: “I ask the world — what is this thing called love, that binds two lives unto death?” Whether through changes of regime or the voluntary exodus west of the pass, history saw continual migration, movement, and fusion of populations between the region north of Yanmen and the steppe beyond the frontier. After the founding of the People’s Republic, and again after the launch of Reform and Opening (改革开放), such migration continued — for example, the construction of industrial facilities in Baotou (包头) and the opening of farmland in parts of Inner Mongolia drew many migrants beyond the frontier.

The richest remains of the region’s role as a frontier garrison date from the Ming dynasty (明朝). The earthen ramparts of the Ming Great Wall still stand on ridgelines and in fields, and many local place names derive from Ming-era military and frontier-defense considerations. Indeed, the Ming was the last dynasty in which the Yan-Yun Zone served as the Central Plains’ bulwark against armed forces from beyond the frontier, facing the constant and formidable military threat of the Oirat (瓦剌) tribes. The Tumu Crisis (土木堡之变), the pivotal event that turned the Ming from prosperity toward decline, occurred on the return journey to the capital after the Ming emperor Zhu Qizhen (朱祁镇) personally led a campaign to Datong. Under the Qing, the empire’s territory expanded vastly northward and westward, and the Yan-Yun Zone’s role as the frontier between China and the lands beyond disappeared forever.

III. Arms and the Great Wall: Defense and Resistance

Xinzhou, the prefecture-level city with the largest territory and the greatest east-west span in Shanxi, is crisscrossed by mountains and marked by diverse landforms. It contains Mount Wutai (五台山), the “roof of North China,” and the headwaters of the Fen, Sanggan (桑干河), and Hutuo rivers — the “three river sources of North China” — while bearing the military burden of “collaring the Central Plains to the south and commanding the great desert to the north.” Separated from Yunzhou and Shuozhou by the Great Wall to the north, bounded by the Shiling Pass (石岭关) toward Taiyuan in the south, girdled by the Yellow River facing Shaanxi and Inner Mongolia in the west, and adjoining the Taihang Mountains (太行山) toward Beijing and Hebei (河北) in the east, it has long been a strategic hub of transport and defense between the Han lands and the northern pastoral world. Thirteen of Xinzhou’s fourteen districts and counties contain sections of the Great Wall, making it the linchpin of the defensive system of northern Shanxi and indeed the Central Plains. According to incomplete statistics, more than 1,700 battles were fought at Yanmen Pass alone.

Some scholars, based on the Tale of King Mu, Son of Heaven (穆天子传), surmise that King Mu of Zhou’s western tour passed north through Yanmen from within Shanxi toward Baotou and the Helan Mountains. Communications between the Han and the Xiongnu, and between the Tang (唐) and the Türks (突厥), likewise ran through Yanmen. In fact, however, the “Yanmen Mountain” (雁门山) of pre-Jin times lay north of today’s Yanggao County in Datong, within present-day Inner Mongolia, and formed the front line of the state of Zhao’s northern defenses during the Warring States period (战国). Thereafter the boundary between the Central Plains and the northern nomads shifted southward several times. Today’s Yanmen Pass was established in the Wei-Jin period and took its name from its proximity to the seat of Yanmen Commandery. In the Tang it was called Xixing Pass (西陉关); in the Song (宋), the Dongxing Pass (东陉关), five kilometers to the northeast, was renamed Yanmen Pass. After the rise of the Liao state, Yanmen Pass became a key garrison against Liao incursions — the stage on which the tragic, heroic saga of the Yang Family Generals, loyal defenders of home and country, was played out. The Ming “Nine Frontier defense system” (九边防御体系), directed against the Oirats and Tatars (鞑靼), placed northern Shanxi’s frontier defenses along the Datong line. The Great Wall system running through Jizhou (蓟州), Xuanfu (宣府), Datong, and Yulin (榆林) essentially shaped the later provincial boundaries between Inner Mongolia and Hebei, Shanxi, and Shaanxi. This east-west Wall system also constituted a “corridor” of culture, history, defense, and trade stretching from Liaodong (辽东) in the east to northern Shaanxi in the west.

After the Yuan dynasty, Yanmen Pass’s character as a military frontier diminished; it became, more importantly, a channel of trade and a commercial entrepôt. Today’s Yanmen Pass lies twenty kilometers northwest of the Dai County (代县) seat. At the “Houyaopu Post Station” (后腰铺驿站) parking lot at the scenic area’s northern entrance, one can still faintly imagine the bustle of frontier trade in former days. Along the road to the pass entrance stands a long row of statues of generals on warhorses, in armor, bearing all manner of weapons — the “Yang Family Generals sculpture group,” twenty-four figures in all, underscoring Yanmen Pass’s historic role as the Central Plains’ bulwark against steppe regimes. In fact, Yang Jiye (杨继业, also known as Yang Ye 杨业) defended the Yanmen region against the Liao as early as the Northern Han (北汉) period. Across the Northern Han and Song dynasties, the Yang family guarded the Yanmen frontier for decades with distinguished service. In nearby Lutijian Village (鹿蹄涧村) stands the Yang ancestral hall, more than a thousand years old; some five hundred descendants of the Yang family live in the village. Although the Yang commanders recorded in the official histories are chiefly Yang Ye and Yang Yanzhao (杨延昭), the Yang Family Generals saga carries rich meanings of history, culture, nation, and state; the stories of Yang Jiye, the “Seven Sons and Eight Tigers,” She Taijun (佘太君), and the Yang family’s women warriors endure in popular literature, the performing arts, and historical research.

From Datong through Yanmen Pass to the vicinity of Taiyuan runs a striking “corridor” — a strategic axis of attack and defense in wartime, and a vital artery of trade and culture in peace. North of Yanmen Pass lies the Datong Basin (大同盆地); to the south, the Xinding Basin (忻定盆地); the two basins are separated by the Heng Mountains (恒山). Mount Heng joins the Taihang range to the east and the Lüliang Mountains (吕梁山) to the west, forming a formidable barrier; only near Yanmen Pass do the mountains narrow and drop in elevation, making it the indispensable defile for north-south passage. South of the pass one enters the Xinding Basin, the gateway to Taiyuan: were the pass lost, Taiyuan would surely fall. Hence Yanmen’s traditional epithet: “Without, it strengthens Datong’s outer shield; within, it secures the lock and key of Taiyuan; it is the root of the Three Passes and the throat of all Shanxi.” The Ming Great Wall, twelve hundred kilometers long and divided into nine garrisons, has Yanmen Pass gripping the juncture of the Heng and Lüliang ranges: eastward it connects through Yanmen to the Inner Wall, westward it forms the “Outer Three Passes” (外三关) with Ningwu Pass (宁武关) and Pian Pass (偏关). Yanmen holds both the strategic significance of being “strategically unmatched” and the cultural-historic standing of being “honored above all.” Reaching Yanmen’s barbican, our study group found, inlaid on either side of the gate, the couplet in the hand of Fu Shan (傅山) — the Shanxi-born thinker, calligrapher, and physician: “Of the three frontiers, a strategic point without equal; of the nine fortresses, the most honored pass of all.”

During the Jiajing reign (嘉靖) of the Ming, Yanmen Pass was extensively enlarged, with a five-kilometer stone enceinte built beyond the fortress city. The gate-tower of this enceinte, the “Bright Moon Tower” (明月楼), now serves as the scenic area’s ticket gate. On site, what one sees most often are the signs and symbols of defense and war. To make assaults on the gates harder, for instance, the barbican gate opens eastward, and steep slopes and sharp bends lie between the barbican and the main fortress, so that once enemies entered the barbican, the defenders could counterattack with battering rams and other weapons. Past the first gate, the Dili (”Earthly Advantage”) Gate (地利门), the road on the left leads to the second gate, the Tianxian (”Natural Barrier”) Gate (天险门), while a “false path” branches to the right — designed both to confuse an enemy who had passed the Dili Gate, unsure which route to take, and to serve as a passage for defenders to encircle the enemy from behind. Atop the pass stand the Temple of Guan Yu (关帝庙), the shrine of Yang Liulang (杨六郎祠), and the shrine of Li Mu (李牧祠); siege engines such as battering carts and gate-blocking blade carts are displayed there today, drawing visitors into endless imaginings of war. The Tianxian Gate is grander and more precipitous still. Entering from the north and passing through it, one’s feet seem to stand upon the soil of the Central Plains. Though ridge upon ridge still crowds the near view, the distant prospect seems already the hills and field-rows of the heartland, stirring an unbidden sense of safety at being within China proper.

An important explanatory variable in contemporary social science for the vast divergence between the historical trajectories of China and Western Europe is the difference between their interstate systems. This so-called Great Divergence (大分流) comparison generally has two “dependent variables”: the modern capitalist economic model and elective parliamentary politics. Why did both emerge in Europe and not in China? The standard explanation holds that China formed a unified state from the Qin and Han (秦汉) onward, and the absence of interstate competition failed to propel breakthroughs in technology and political institutions; Europe, by contrast, developed a system of many competing states, whose fierce rivalry drove not only the Industrial Revolution and modern science and technology but also the modern military-fiscal state. Yet standing on the front line of contest between the Central Plains and the steppe — in the Yan-Yun Zone — we could not help but reconsider this question. Although China was unified in the Qin-Han era, East Asia also experienced long periods of intense interstate competition. Leave aside the Wei-Jin period, when northern nomadic tribes surged southward, squeezing the Jin dynasty (晋朝) south of the Yangtze and creating in North China a multistate system of sixteen successive polities; even the mighty Tang endured massive shocks from the Tibetan Empire (吐蕃), and during the An Lushan Rebellion (安史之乱) the capital Chang’an (长安) briefly fell. During the two Song dynasties alone, the East Asian continent contained, in succession, the Song, Western Xia, Liao, Tibetan polities, Jin, and Yuan; and the military competition among Song, Western Xia, Liao, Jin, and Yuan repeatedly ended in the annihilation of states — competition that can hardly be called anything but fierce. Why, then, did such interstate rivalry produce nothing like Europe’s great transformation — the great voyages, the opening (in truth, plundering) of the Americas, the colonization of all Africa and nearly all Asia — nor a gunpowder-and-firearms revolution in warfare that would spur physics and chemistry and usher in a modern scientific revolution?

Part of the answer lies in the moral constraints the Eastern Confucian-Legalist state (儒法国家) imposed on the polity: its military development was always oriented toward self-defense, and it never treated foreign lands as objects of plunder, occupation, colonization, or enslavement. Under such conditions, when land and resources ran short, there was spontaneous popular emigration — “going west of the pass” or “sailing south to the Nanyang” (下南洋) — but state-sponsored expeditions of exploration and plunder were impossible. The utterly different aims and consequences of Zheng He’s (郑和) voyages to the Western Oceans, on the one hand, and Vasco da Gama’s rounding of the Cape of Good Hope or Columbus’s landfall in the Americas, on the other, are proof of this. Another part of the answer is that the Confucian-Legalist state, grounded in an agrarian economy, displayed enormous resilience in the face of nomadic attack: the sheer mass of the Chinese agrarian socioeconomic system gradually wore down external forces. In short, even if the territory south of the Great Wall fell, a Central Plains dynasty could retreat south of the Yangtze and survive, even flourish (as the Southern Song 南宋 and Eastern Jin 东晋 both did over long periods), while nomadic regimes, even after conquering North China, could never fully occupy or colonize it — instead they were absorbed into, or actively converted themselves to, the agrarian socioeconomic model. From the long view of history, that is, relative to the enormous mass of China’s agrarian socioeconomic system and Confucian-Legalist political order, the interstate competition that did occur remained a “storm in a teacup.”

Following the natural boundary between the semi-arid and semi-humid zones, and guided by the principle of “following the terrain and using natural barriers to control the frontier,” the Central Plains dynasties built fortified passes at mountain defiles and river bends. The Yanmen section of the Great Wall is bound lip-and-teeth to Gouzhu Mountain (句注山). Part of the western Heng range, Gouzhu runs roughly southwest-northeast at elevations of about 1,700 to 2,400 meters — commandingly steep yet still climbable, and thus well suited to defensive works. Of course, owing to climatic differences across eras, the Wall’s location could vary greatly. Comparing the Qin Wall with the Ming Wall, the latter’s southward retreat is plain to see, with a range of fluctuation between 200 and 400 kilometers. Moreover, the dynasties built walls of differing thickness according to local risk levels. Walls on high summits often had two or even several courses; flat terrain near the political center received the heaviest protection of all — the Hetao Plain (河套平原), or the Juyong Pass (居庸关) and Shanhai Pass (山海关) areas close to Beijing, had as many as twenty layers of wall. Jia Yi (贾谊) of the Western Han, in his “Disquisition on the Faults of Qin” (过秦论), wrote that Meng Tian’s wall “drove the Xiongnu back more than seven hundred li, so that the nomads dared not come south to graze their horses.” Yet in later ages the Liao, Jin, Yuan, and Later Jin (后金, i.e., Qing) — northern regimes of minority peoples — all successfully crossed the Great Wall, occupying vast territories and even all of China. With the consolidation of the Qing, a dynasty of steppe character, this institutional feature of the Great Wall vanished, and the Chinese nation and Chinese civilization acquired a broader meaning.

In the twentieth century, however, the Great Wall once again saw the fires of war. After occupying the Northeast in 1931, the Japanese army quickly plotted to push south of the Wall, attacking Rehe (热河) and the Wall line east of Gubeikou (古北口) in 1933. China organized the Defense of the Great Wall (长城抗战), fighting the Japanese at Shanhai Pass, Xifengkou (喜峰口), Gubeikou, and elsewhere. Apart from the pyrrhic “Victory at Xifengkou” (of five hundred broadsword-brigade soldiers in that battle, only twenty-three survived), every engagement was lost, and China was forced to sign the Tanggu Truce (塘沽协定). North China’s gate now stood open; Japan pressed forward with plans for full-scale invasion, until in 1937 it launched the Marco Polo Bridge Incident (卢沟桥事变) and occupied Beiping (北平, now Beijing).

Against the backdrop of twentieth-century warfare — firearms and mechanized armies — the Great Wall’s function as a military installation had essentially been lost, but its institutional character as a political and identity boundary grew all the more salient. Once the Japanese forces already holding the three northeastern provinces breached this line, the danger of China’s total subjugation had begun. The bitter resistance waged along the Wall in Hebei and Shanxi after 1933 and 1937 must be understood within this framework. Many major battles took place in the Yan-Yun Zone. In October 1937, the Eighth Route Army (八路军) ambushed a Japanese transport column on the highway south of Yanmen Pass — the Yanmen Pass Ambush (雁门关伏击战). The same month, the Eighth Route Army staged a night raid on the Japanese airfield at Yangmingbu (阳明堡) in Dai County, killing more than a hundred Japanese troops and destroying or damaging twenty-four aircraft. Pingxing Pass is famous for the Eighth Route Army’s major victory there. In September 1937, a Japanese detachment attacking Shanxi approached the pass; the 115th Division of the Eighth Route Army laid an ambush on favorable ground, ultimately killing more than a thousand enemy troops and capturing large quantities of weapons and matériel — the Great Victory at Pingxingguan (平型关大捷). Yet the invading Japanese army eventually took Xinkou (忻口) and Taiyuan, and the Chinese people’s War of Resistance would not be won for another eight years. Yanmen Pass and other Wall defenses were gravely damaged during the Japanese assault and occupation. A local propaganda-department colleague told us that all the timber of the buildings of the Zhenbian Shrine (镇边祠) at Yanmen Pass was burned by Japanese troops for heat; only two pines in the courtyard survive, still green and towering today.

Pingxing Pass lies on the border of Fanshi and Lingqiu (灵丘) counties, named for its bottle-shaped terrain. With Mount Heng rising like a screen to the north and Mount Wutai towering to the south, it is the ideal pass for entering Shanxi from Hebei. In 1511, the sixth year of the Zhengde reign (正德) of the Ming, the Lingkou Fort (岭口堡) was built, making Pingxing Pass a key gate of the Inner Great Wall. Linked eastward to Zijing Pass (紫荆关) and westward to Yanmen Pass, it formed part of a tight defensive line. Pingxing Pass still crouches tiger-like over the defile today, comprising its Wall section, fortress city, fort (now Pingxingguan Village 平型关村), old pass road, beacon towers, drill ground, and the Liulang Fortress (六郎城). The old Wall was built to exploit natural hazards: a city at the pass, walls joined to the ramparts, forts beyond the walls, and beacon towers standing guard to relay alarms. The fort nearly three kilometers southwest of the fortress city — today’s Pingxingguan Village — still preserves its barbican and gate intact. Within, the houses stand in neat order; in former times it housed hereditary military households, with civil and military yamens, the sizable Sanjiao Temple (三教寺), Bixia Temple (碧霞寺), and City God Temple (城隍庙), and many shops. At the village entrance stands a chastity arch commemorating a Qing-era woman née Zhang: after her husband died at his post in Henan (河南), she kept her widowhood for sixty-one years from the age of twenty-one, serving her parents-in-law, treating her neighbors with kindness, and raising her son through hardship until he won official rank. Clearly, even in a frontier fort like those of the Yan-Yun Zone, the original inhabitants might come from distant provinces of the interior, and over sufficient time they gradually formed an organic cultural community. Walking through the Yan-Yun region, one passes such forts again and again.

In the lanes of these forts, walking past each family’s windows and under their eaves, our study group could not help wondering: when, and from where, did the ancestors of the family inside migrate here and settle? After all, state-directed migration from the interior to frontiers and strategic strongholds has been an important policy option for securing borders in both ancient and modern states.

IV. Pass Markets and Trade

Fei Xiaotong (费孝通) held that although the Central Plains and the north were constantly at war throughout history, the more important side of the story was the regular, mutually dependent exchange and trade that went unrecorded. In times without war, military strongholds such as Yanmen Pass and Shahukou became hubs of “pass market” (关市) trade. The Shanxi merchants (晋商), having accumulated considerable capital in the early and middle Ming through the kaizhong system (开中制) — delivering grain to the frontier in exchange for salt certificates — and knowing the frontier garrisons and trade routes intimately, rose swiftly once the border markets opened. Based in several gateway towns, they stockpiled horses, mules, and sheep for resale to frontier garrisons or the interior, while selling Central Plains handicrafts and agricultural goods to the frontier regions; trade in both directions yielded astonishing profits. It was said that a single iron cauldron carried out through Yanmen Pass could be exchanged for a horse, which was then sold at a profit in the interior.

But the goods passing through these gates were by no means limited to the products of the two sides of the Wall. In fact, this commerce was part of trade across the whole of Eurasia, shaping the continent’s economic corridors. In the history of world trade, besides the Silk Road (丝绸之路), routes named for their principal goods include the Spice Route, the Jade Road (玉石之路), and the Tea Road (茶叶之路); the transport networks for salt, iron, and other goods were also very dense. Historically, tea set out from the Wuyi Mountains (武夷山) of Fujian (福建), passing through Jiangxi, Hubei, Henan, Shanxi, Hebei, and Mongolia, relayed by porters, boats, horse caravans, and camel trains, by water and by land, to be traded at Kyakhta (恰克图), whence Russian merchants carried it to Moscow and St. Petersburg and into Europe. Stretching more than ten thousand kilometers from Wuyishan to Europe, this route was called the Ten-Thousand-Li Tea Road (万里茶道). Yanmen Pass held the economic lifeline of Daizhou, Guo County (崞县), Wutai (五台), Dingxiang (定襄), Xinzhou, and Fanshi to the south, and Ying County (应县), Shuo County (朔县), Shanyin (山阴), Hunyuan (浑源), and even Datong to the north. Standing today at the front and rear Yaopu post stations and the frontier-trade street within the Yanmen Pass scenic area, one can still dimly picture the prosperity of that commerce. Descending from the Tianxian Gate, the mountain road twists and turns, steep in places. Looking at the deep cart ruts worn into the flagstones and exposed bedrock, one can imagine camels, mules, and horses hauling loaded carts laboriously up the mountain.

The village built against the mountainside within the walls of Yanmen Pass was historically called Yanmen Stockade (雁门寨), renamed Mingyue Village (明月村) in the Qing and Yanmenguan Village (雁门关村) in the early Republic — a typical “frontier-pass village.” Its villagers were, historically, frontier people through and through. Those inside the pass called them people from beyond it; in the eyes of those beyond, they were people from within. Neither southern nor northern, they nonetheless connected south and north, fusing inside and outside. They were the human embodiment of the junction between the Central Plains and the lands beyond, and to this day their way of life and customs bear clear marks of blending north and south, Hu (胡) and Han (汉). Their chief livelihood was helping merchants haul goods up rugged mountain paths with a thousand meters of elevation change — a profession called “guarding the pass” (护关), which faded only after Yan Xishan (阎锡山) cut roads through the mountains for motor traffic in the early Republican era.

The Central Plains regarded this region as a frontier, or as a transition zone between agrarian and nomadic civilizations. In the imagination of people from the heartland, the frontier ought to be a bitter, desolate place, its two sides sealed off from each other — if not in constant military conflict, then at least mutually wary and hostile. Most of the time, however, local life was nothing of the sort. The “Account of the Xiongnu” in the Book of Han (汉书·匈奴传) already described the lively traffic on both sides with the phrase “coming and going beneath the Great Wall.” Geographic constraints made it hard to establish complete Central Plains or nomadic institutions in the transition zone, and loose social control gave frontier people greater room for maneuver in the distribution of gains. Gaps in corvée and taxation and the drifting of laboring populations produced a flexible, free market economy — while also intensifying collusion between officials and merchants, and between the military and merchants. Traders, adept at profiting from the needs of army camps, often gained from military conflict. The latent military risks and banditry of the border region deepened residents’ dependence on fortresses: dwelling, labor, trade, and business all centered on the various military stockades and forts — perhaps the most direct relationship between local residents and state power.

Constrained by natural ecology and economic structure, the nomadic tribes had enormous demand for the tea, salt, silk, ironware, and tools of the lands south of the Wall, while the Central Plains had great demand for horses from beyond it. Horses were vital for farming, transport, and war — so why did the vast and abundant Celestial Empire never develop large-scale horse pastoralism south of the Wall? Given its productive efficiency, the reliability of its food supply, and its potential for population growth, agriculture was the best avenue of food provision when productivity was still limited. In the Tang era, one square kilometer of farmland could sustain 62.5 people, while the same area of pasture could support only 6. In arid and semi-arid zones, environmental conditions severely restricted crop farming, compelling people to turn to pastoral production, converting grass into nutrition through cattle, sheep, and horses to secure subsistence — an effective social model formed in conformity with natural geography. Agrarian civilization in the same period achieved population growth through labor-intensive farming. From the Spring and Autumn and Warring States periods (春秋战国) onward, agriculture in the middle and lower Yellow River valley shifted steadily toward intensification, with yields often exceeding 50 kilograms per mu (亩) — clear evidence that per-unit grain yields had reached scale, and population grew rapidly. As the Song economy’s center of gravity moved south, northern migrants used two techniques — rice transplanting and rice-wheat double cropping — to raise land-utilization rates in Jiangnan (江南) from 50% to 200%. Ge Jianxiong (葛剑雄) has pointed out that while agricultural production is not the sole cause of population growth, every major breakthrough in China’s demographic history was the direct result of increased grain output; no demographic miracle could arise except on the foundation of a great expansion of agriculture. Thus, with the spatial extension of farming and yield gains from technical inputs, China’s population surpassed 200 million in the early-to-mid Ming and reached a new peak of 430 million by the mid-nineteenth century.

In most cases, Central Plains society could not convert cropland into pasture for stock-raising, so horses had to be imported from the north. Horse markets were established at many Datong forts — Shoukoubu, Zhumabu (助马堡), Bao’anbu (保安堡), Ninglubu (宁虏堡), Shahubu (杀胡堡), Yunshibu (云石堡), Ying’enbu (迎恩堡), Miehubu (灭胡堡), and others. As for the tea markets: from the Tang and Song onward, Central Plains dynasties generally held that tea was indispensable to the pastoral “barbarians” who followed water and grass. The History of Ming (明史), chapter 56 (”Food and Money IV”), records: “The tribespeople relish milk and cheese; deprived of tea, they suffer and fall ill.” From this arose even a political doctrine of “controlling the barbarians through tea” (以茶制夷), and with it the institution of tea-for-horse border markets. The Ming rulers in particular, as noted above, hoped to use the abundant products and advanced culture of the Central Plains to win over, divide, and even control the nomadic regimes, inducing their sincere submission and thus removing the frontier threat: horse markets were set up in the east to bridle the Mongols and Jurchens, and tea markets in the west to pacify the Tibetans. Tea was yet another embodiment of agro-pastoral fusion: the nomadic peoples of the old frontiers universally drank tea, developing distinctive “border tea cultures” (边茶文化) blended with local and ethnic characteristics.

Douglass North and Robert Thomas argued that in the eleventh and twelfth centuries, population pressure drove residents across Europe to migrate to more remote, previously uncultivated areas. The formation of new settlements meant differences in products between regions, creating the conditions for trade; hence cities gradually arose at the crossroads of trade routes. On the East Asian continent, the economic modes and products on either side of the 400mm isohyet differed markedly, generating enormous trade potential. If North and Thomas are right, then trading metropolises like Europe’s Venice and Bruges could well have emerged along this boundary. Dunhuang (敦煌), Zhangjiakou (张家口), Shahukou, and Yanmen Pass were all once flourishing trade centers; Quanzhou (泉州) in the Song-Yuan era and Guangzhou (广州) in the Ming-Qing era became thriving trading cities as hubs between China, Southeast Asia, and the Indian Ocean, and Southeast Asia itself produced many trading port cities or “port states” of over a hundred thousand inhabitants. Had Zhangjiakou, Yanmen Pass, and Shahukou (or Datong, Dai County, and Xinzhou) developed into trading cities like Venice or Bruges rather than military strongholds, might China’s economic history have been rewritten — and might later economists like North have studied the rise of China or Asia rather than of Europe?

This touches another great debate in scholarship: why commercial economy never rose on a grand scale in China, and why capitalist industrialization did not occur there. Zhao Dingxin argues that two features of China’s Confucian-Legalist polity were decisive: a unified Confucian moral order restrained members of society from pursuing commercial gain, and the merchant class could profit only by attaching itself to officials (the state), never wielding significant economic, military, or ideological power. Wen Yi (文一) argues that traditional China’s moral order kept it from the plunder, massacre, and colonization of the Americas, Africa, and Asia that marked Europe’s Age of Discovery — and that it was precisely those acts that supplied the wealth-foundation of Western Europe’s Industrial Revolution. Indeed, a polity staffed by Confucian scholar-officials could never tolerate the emergence, in frontier regions, of large trading ports or cities beyond official control; commercial development was thus strictly confined within narrow limits. More broadly, the Central Plains dynasties regarded the lands beyond the Great Wall as “outside the realm” — at once tributaries and potential threats to border security, even to the capital and the regime itself. Under such conditions, the frontier’s military role wholly suppressed the possibility of trading cities arising.

Even so, “mutual markets” (互市) of considerable scale did form along the frontier; military strongholds like Yanmen Pass and Shahukou became centers of border trade, and, more, became the historical memory of two cultural spheres coexisting, communicating, and blending. The Ming jingshi wenbian (明经世文编) records a memorial by Wei Shiliang (魏时亮) on Gansu: “The tribes cannot go unsuccored; the tribespeople are our people. Our lands adjoin like lips and teeth; their weal and woe are one with ours.” Plainly, farmer and herder developed an intimate, mutually generative relationship — even sentiments of shared honor and hardship, of common cause. Locals told us that the people of Dai County have many relatives in Datong and Inner Mongolia — a typical example of ethnic fusion. In the interweaving of nomadic and agrarian life, intermarriage was an inevitable fact. Marriage opened a window onto deep knowledge of the other’s language, culture, religion, and customs, and even altered the gene pool. (The so-called fujia (复甲) hypothesis may serve as evidence.) In the agro-pastoral transition zone one comes to understand profoundly that the Hu Huanyong Line is not merely a boundary of population density but a watershed between two types of culture. It was precisely through the contact, exchange, and fusion of nomadic and agrarian civilizations that the grand architecture of Chinese civilization was built up layer by layer. As the saying goes, “One Yanmen Pass, half the history of China” (一座雁门关，半部华夏史): the emperors who toured here, the generals who fought here, the princesses who passed through on marriage-alliance journeys, and the soldiers and civilians who kept the pass together created the cultural symbol of Yanmen — through which later generations may glimpse, as through a keyhole, the splendor of civilizations north and south of the Wall, so different in temperament yet so deeply mutually constitutive.

At Yanmen Pass under a full moon, walking the mountain road in darkness, the bright moon in the night sky calls to mind the poem “Mountain Studio Shaded from the Moon” (蔽月山房), written by the young Wang Shouren (王守仁, Wang Yangming 王阳明): “The mountain near, the moon far off — the moon seems small; so people say this mountain is bigger than the moon. But were there one whose eye was vast as heaven, he would see the mountain high — and the moon wider still.”

V. Geography and Human Society in the Industrial Age

The 400mm isohyet demarcated different zones of economic production, shaping the divide between agrarian and nomadic civilization. The influence of “climate and soil” on economic forms — Ray Huang’s abiding preoccupation — further shaped the state’s political-economic system, producing what he called the “precocity” and “ultra-stable structure” of Chinese politics, which blocked an industrial revolution. Zhao Dingxin argues that the Confucian-Legalist state’s near-perfect fusion of ideological and political power made it impossible for the phenomena of early modern Western Europe — civil society, religious reformation, an urban commercial class, industrial revolution — to appear in China. Wen Yi’s explanation of the Western industrial and scientific revolutions stresses instead the predatory cultural tradition formed in the West after the Crusades, and the competitive state system’s crushing demand for munitions and for the fiscal extraction to sustain war, generating a cycle in which war fed capitalism and technological revolution and vice versa, ultimately producing the West’s industrial revolution and modern scientific revolution. The significance of the 400mm isohyet, however, lies in its direct influence on productive activities and organizational forms tied to land and vegetation. Once human society enters the industrial age, the influence of rainfall on socioeconomic development declines; in some cases the effect of climate and rainfall falls almost to zero — the oil states of the Middle East being an example.

There are three mechanisms by which the influence of climate and natural geography declines in the industrial age. First, the structure of resource endowments changes: rainfall and water are no longer the decisive endowments, and minerals become a decisive resource. Second, the mechanisms driving economic growth change: human resources (labor, knowledge, and technology) and capital become paramount. Third, as industrialization advances, humanity’s capacity to actively transform and adapt to the natural environment improves. Irrigation systems, water-extraction and storage systems — including Kuwait’s seawater desalination — river regulation, and urban greening have all vastly enhanced humanity’s ability to cope with natural geography. And improved capacity to build transport infrastructure has powerfully offset the constraints that physical distance across mountains, deserts, and other terrain once imposed between peripheral regions and economic heartlands. The construction, after 1949, of industrial center cities such as Baotou and Datong north of the 400mm isohyet exemplifies these mechanisms at work.

The Yan-Yun Zone, and Shanxi as a whole, are quintessential cases. Since the modern era, Shanxi’s industrialization long ranked among the most advanced of China’s provinces, with substantial development in steel, armaments, cement, chemicals, and light industry. In the early years of the People’s Republic, Shanxi’s degree of industrialization was at one point second in the nation, behind only Liaoning (辽宁), then the great industrial province. Yet the famous “resource curse” of political science and economics seems also to have operated in Shanxi: in the latter half of the twentieth century, the province developed an economic structure dominated by the coal industry, and after the turn of the twenty-first century the many negative consequences of this model became increasingly evident. Along our route, our study group still saw abundant coal-mining facilities, and the railways and highways still visibly bore the burden of coal transport. Local colleagues frequently discussed the enormous pressure Shanxi faces to restructure its economy; in developing innovative sectors and electronics and information industries, Shanxi lags far behind the eastern provinces. In fact, however, Shanxi is pressing actively ahead with high-quality development in industry and informatization.

If the 400mm isohyet’s power to divide economic and human geography has weakened so greatly in the industrial age — and if the state has actively organized migration and development across the line, as with the Xinjiang Production and Construction Corps (新疆生产建设兵团), and industrialization and urbanization projects such as the rise of Baotou and Karamay (克拉玛依) — why does the Hu Huanyong Line still function so conspicuously as a dividing line in China’s economic and human geography today? On today’s economic and demographic map of China, the two sides of the line still differ as heaven from earth in population density, in the density of expressway and high-speed rail networks, in per-capita electricity consumption, and in daily mobile-data traffic. One reason is classic path dependence: the southeastern side of the Hu Huanyong Line was already the center of gravity of China’s population and economy — an established fact — when the country embarked on socioeconomic modernization, and this fact is deeply inscribed in the trajectories of economic and demographic geography during modernization. Another reason is that in other endowments affecting economic and demographic patterns — such as proximity to international markets and international capital — the southeastern side also holds clear advantages. The Hu Huanyong Line, in other words, is not merely a line dividing rainfall; it divides other endowments as well.

In any case, across different historical periods, many state-led efforts to promote national development and especially regional balance have greatly ameliorated regional disparities: the Third Front construction (三线建设), the Western Development program (西部大开发), targeted poverty alleviation (精准扶贫), paired assistance between eastern and western provinces (对口支援), the westward relocation of textile spindles (东锭西移), West-to-East power transmission (西电东送), the Belt and Road Initiative (一带一路), the East-Data-West-Computing project (东数西算), and the construction of the “eight vertical and eight horizontal” high-speed rail network and ultra-high-voltage transmission grids. In the Yan-Yun Zone we saw large arrays of photovoltaic installations built across farmland. In a region short of water but rich in sunshine, the opportunity costs of cultivation and solar generation clearly diverge enormously. Yet converting solar resources into economic income became possible only in recent years, once the state had provided power-transmission infrastructure and built a well-functioning electricity market. Our study group also saw the Xiong’an-Xinzhou high-speed railway (雄忻高铁) under construction, linking the Yan-Yun Zone to the economic core of North China and heralding the zone’s imminent integration into the one-hour transport circle of Beijing, Tianjin, and Xiong’an (雄安) — another example of how the industrial age transforms the structural constraints of natural geography and powerfully reshapes the economic map.

Over the past two decades, China’s nationwide efforts to improve the human living environment have drawn worldwide attention, and the Yan-Yun Zone’s achievements are striking. More than half of Xinzhou’s counties lie in the gully-scarred loess hill region, suffering severe soil erosion and frequent drought. In recent years, through combined efforts in water conservancy, agronomy, and mechanization, the fields are now level and fertile, irrigation works complete, and township road and shelterbelt networks unimpeded. To address water scarcity, small watersheds were tackled one by one — securing water supply, comprehensive management, ecological cleanup, and reinforcement of county-level small water networks — gradually raising the utilization rate of natural precipitation, strengthening resilience against natural disasters, and building the backbone of a modern water network of considerable scale. According to a report of the Xinzhou Municipal Government, in 2021 alone the city invested 501 million yuan in 203 key water projects across eight categories. Data show that Xinzhou’s green-coverage rate has risen from under 4% in the 1980s to 40% today, with urban per-capita park green space reaching 14.89 square meters. Traveling north from Taiyuan through Fanshi, Dai County, Datong, and on to Ulanqab, our study group could feel the effects of this ecological improvement everywhere.

And yet China’s efforts to improve its domestic ecology and living conditions may count for little against the backdrop of global climate change. Industrialization has raised humanity’s ability to transform local environments, but the long accumulation of carbon emissions has set the planet’s overall environment on a warming trajectory. China has acted vigorously in the global response to climate change, advancing toward carbon peaking by 2030 and carbon neutrality by 2060. But judging from current progress, humanity faces the challenge of failing to hold global temperature rise within 2°C this century. Global warming will not only devastate biological systems but greatly increase the scale and frequency of extreme weather — heat waves, blizzards, and the like. Rising temperatures are melting polar ice and the snowfields and glaciers of the world’s mountains, with sea-level rise as the direct consequence: not only will many island nations be entirely submerged, but the coastal regions of continental countries — such as China’s Yangtze River Delta (长三角) — will also be inundated. Paradoxically, for the long-arid, long-cold Yan-Yun Zone and the regions north of the 400mm isohyet, climate change is good news: these areas will receive more rainfall and a relatively warmer climate. Perhaps, over the next twenty to thirty years, as the eastern seaboard gradually faces the prospect of inundation, ever more Chinese will migrate northwest — to the Yan-Yun Zone, the Hetao region, Inner Mongolia — to live and work? As a country of vast territory, we may have living space enough to weather the catastrophic prospect of global warming; many other countries in the world, one fears, will have no such luxury.

When our study group left Taiyuan for Xinzhou, heavy rain fell on Taiyuan. Encountering a downpour in a historically arid region, one could not help recalling the torrential rains and floods that have struck Henan, Shandong (山东), and Beijing in recent years. The rain band then followed our footsteps northward; the night we reached Datong, the rain clouds arrived with us. Looking out the window the next morning, the rain was drumming down; water had pooled on the ground below and on the rooftops opposite; parents ferrying children to school on electric scooters all wore rain capes — a scene one expects of the Jiangnan rainy season. To meet such rainfall north of the traditional 400mm isohyet seemed to signal that the astonishing prospect of global warming is already knocking gently at the door.

Conclusion

On grand social-scientific questions — the interaction between agrarian and nomadic societies, and the influence of geographic conditions on economic and social patterns — this article has cross-validated and re-examined existing multidisciplinary theory against field observation. Its central concern has been the Hu Huanyong Line, or 400mm isohyet, which has so profoundly shaped Chinese history and China’s socioeconomic development. From it we see that natural geography exerts a deep influence on long-run economic, social, and political patterns.

In closing, we must answer one more question: if rainfall has declined in importance in industrial society, what is the point of studying its influence on socioeconomic patterns today? In fact, although rainfall matters less now, the Hu Huanyong Line’s influence on China and East Asia, accumulated over millennia, will persist for a long time. Simply put, the economic and demographic geography of China formed under the line’s influence over thousands of years will not change significantly in the short run. Related to this are, on the one hand, the long and enduring influence of history and culture on human society, continually narrated and transmitted; and on the other, the new meanings people will find within new socioeconomic patterns amid rapid modernization. In the four-plus decades since Reform and Opening especially, Chinese society has undergone, within a single generation, changes that took Western societies two or three centuries. As socioeconomic patterns transform at speed, people’s ways of life and the physical spaces and geographic environments they inhabit are also changing enormously. How, amid such upheaval, to understand and accept new ways of living and new social realities while maintaining psychological and cultural ties to environments and social worlds that are past — perhaps gone forever — is the great challenge that a rapidly changing society poses to the individual. The ordering and narration of past history is, one might say, itself a way of coping with change.

In the Yan-Yun Zone we saw the Xiong’an-Xinzhou high-speed railway drawing the region into a new transport “corridor,” solar installations in the fields bringing it into a new power and energy “corridor,” and the growth of Shanxi’s manufacturing and information industries incorporating it into a new industrial “corridor.” All this means that this land, once so constrained by rainfall and soil erosion, is — like many other parts of China — steadily entering a mode of economy centered on industry. This is a manifestation of the great success of Chinese-style modernization. But understanding and embracing the new socioeconomic order will, for many, still take much time. At the very least, this mysterious geographic dividing line, which has shaped the East Asian continent since the dawn of Chinese civilization, remains fascinating. Perhaps, as animals, human beings’ intense feelings for mountains, rivers, valleys, plains, clear skies, and rain and snow are written in our genes. And so today — when high-speed rail and expressways, irrigation and river systems, air conditioning and heating continually improve the human habitat and loosen geography’s grip on social and economic life — it remains deeply meaningful to keep asking what geography means for human activity. Against the backdrop of drastic global climate change, the habitable range of humanity on the earth’s surface may well shift dramatically; many people may have to leave the geographic spaces they once knew and migrate to new ones. If such a future is destined to arrive, then the mutual shaping of geography and human society deserves our sustained attention.

(This article is dedicated to Ray Huang [黄仁宇], who devoted his career to the long-term trends of Chinese history — to the tug-of-war, conflict, and fusion between Central Plains civilization and the nomadic civilization of the north, and to the invasions and shocks endured by China and East Asia after the arrival of capitalism and European imperialism and the rise of modern Japan.)

More to read:

The initiative’s director, Hu Bo, is a prominent scholar in China’s maritime strategy field. He currently serves as a research fellow at Peking University’s School of International Studies and as director of the Center for Maritime Strategy Studies at Peking University. His work centers on maritime strategy and policy, with particular focus on sea power strategy, the South China Sea, international security, and the US military. He is the author of several books, including China’s Maritime Power in the Post-Mahan Era.

Back in January of this year, I published one of their analyses in my newsletter; Mingkun Technology—whose chief scientist is none other than SCSPI director Hu Bo—had detected US forces rapidly redeploying to Europe and the Middle East in the wake of American operations in Venezuela, and used those observations to map out the likely pattern of US action against Iran. I also interviewed Hu Bo on the subject. In his view, Iran’s domestic situation is the key variable; Trump, he argued, is adept at picking low-hanging fruit, inclined to seize the moment when internal turmoil breaks out in Iran, but only when he can guarantee an absolute advantage. If Iran stays internally stable, Washington will struggle to find an opening anytime soon. Subsequent developments, along with the ceasefire the two sides have since reached, have borne out his assessment.

The report argues that the US is transforming the Philippines into a forward hub, which looks more like an asset-light exercise in strategic posturing than actual preparation for war. It notes that since Marcos Jr. came to power in 2022, the Philippines has shifted from a balanced diplomacy to the front lines of confrontation. The number of bases under the Enhanced Defense Cooperation Agreement (EDCA) has jumped from five to nine, forming a strategic layout aimed at China that “locks down the Taiwan Strait to the north and controls the South China Sea to the south.”（北锁台海、南控南海） Advanced systems such as HIMARS, the Typhon mid-range missile launcher, and the NMESIS interceptor have arrived in succession, and the US presence is now said to “amount to a permanent military presence” in the country.

Yet the report also delivers a dose of cold water. These forces, it contends, are largely a peacetime presence and would prove highly vulnerable in an intense conflict; their real value lies in the political, strategic, and diplomatic realm, to disrupt the situation, that is, leveraging the Philippines’ geographic proximity to tie down and wear down China rather than to prepare directly for war. Because the Philippine bases are poorly developed and lack wartime resilience, the US has remained cautious about investing in their expansion, and in strategic terms, the Philippines serves more as a supplement to US bases in Japan than as a frontline in its own right. The report thus arrives at a somewhat ironic conclusion that the Philippines’ hope of using American power to deter China is little more than wishful thinking. For China and other stakeholders, the takeaway is that while the wartime threat of these deployments deserves attention, even greater vigilance should be paid to how they tie down and drain China in peacetime.

Below I’ve attached the brief version of the report. Thanks to the kind authorization from Hu Bo, I can share the report in my newsletter

Full report in English

1.84MB ∙ PDF file

Download

Download

U.S. Military Base Expansion and Force Posture Enhancement in the Philippines

Preface

Since Ferdinand Marcos Jr.’s inauguration in June 2022, the U.S.-Philippine alliance has strengthened significantly. The number of sites under the Enhanced Defense Cooperation Agreement (EDCA) has surged from five to nine, with the site in northern Luzon facing the Taiwan Strait and the site on Balabac Island approaching the frontline of the Nansha Islands (Spratly Islands). Consequently, the U.S. military presence in the Philippines now has acquired the practical significance of a “forward position on the first island chain”. Meanwhile, the scale and intensity of joint exercises such as Balikatan, Salaknib, and KAMANDAG have steadily increased year after year. Systems including the High Mobility Artillery Rocket System (HIMARS), the Typhon Weapon System, and the Navy-Marine Expeditionary Ship Interdiction System (NMESIS) have been successively deployed. Through these developments, the U.S. military is progressively incorporating the Philippines into its Indo-Pacific strategic system and command framework.

Nevertheless, the expansion of EDCA bases has fallen short of expectations, and U.S. efforts to strengthen its military presence in the Philippines continue to face constraints from multiple factors. First, the Philippines’ own capabilities are limited, providing little assistance to the U.S. in wartime; Second, while the U.S. needs the Philippines’ strategic geographic location, it would require significant investment to fully enable the EDCA sites for wartime readiness; Third, the political uncertainty within the Philippines and rising nationalism have added to U.S. hesitations.

Although U.S. forces have sharply increased their presence and deployments in the Philippines, these assets possess limited wartime utility, especially against a high-end opponent such as China. The main purpose of this strengthened military access is not genuine preparation for war, but rather strategic posturing, containment, and attrition: exploiting the Philippines’ geographical advantages to bolster U.S. force posture around the South China Sea and Taiwan Strait, shaping a favorable overall strategic environment, and using the Philippines to wear down China during peacetime. At the same time, the U.S. is cautious about making excessive investments in these sites and is wary of turning the Philippines into a wartime forward base.

This report aims to objectively outline and evaluate the expansion of U.S. military sites and other military presences in the Philippines, analyzing their impact on regional security in both peacetime and wartime for the reader’s reference and consideration.

Director of SCSPI Hu Bo

Key Findings

I. Base Expansion: From Strategic Access to Forward Presence

The expansion of U.S. military bases in the Philippines is a critical link in America’s strategy to implement “Integrated Deterrence” and optimize the Indo-Pacific pattern. The core objective is to transform the Philippines from a “strategic rear” into a “forward hub”. The northern base group is designed to block the Bashi Channel and Balintang Channel during crises and cut off access routes to the Western Pacific; the southern base group strengthens persistent presence and intelligence-collection capabilities in the South China Sea, providing direct support for sea-control operations. In recent years, leveraging the EDCA, the U.S. military has continuously pushed forward infrastructure expansion of military bases in the Philippines to enhance their capacity to support realistic combat-oriented joint drills and maritime cooperation activities.

In March 2016, following the annual Philippines-United States Bilateral Strategic Dialogue, the two governments issued a joint statement announcing that, pursuant to the 2014 EDCA, U.S. forces would be granted access to five Philippine military bases. These sites are Basa Air Base in Pampanga, Fort Magsaysay in Nueva Ecija, Antonio Bautista Air Base in Palawan, Mactan–Benito Ebuen Air Base in Cebu, and Lumbia Air Base in Cagayan de Oro. In February 2023, the U.S. and the Philippines announced plans that in addition to the previous five bases invested and stationed by U.S. forces, they would establish and operate four additional new bases. In April, the Philippines announced the locations of these four additional sites: Naval Base Camilo Osias and Lal-lo Airport in Cagayan, Camp Melchor Dela Cruz in Isabela, and Balabac Island in Palawan.

II. Force Deployment: From Rotational to Combat-Ready Military System

The characteristics of U.S. force deployments in the Philippines have undergone three major shifts: from intermittent exercise-based presence to sustained rotational deployments; from single-service operations to integrated joint operations; and from bilateral cooperation to multilateral alliance coordination. The U.S. military presence in the Philippines has already gone beyond the traditional scope of alliance cooperation, evolving into a fully functional, rapidly responsive, and tightly coordinated, combat-ready military system integrated with multilateral alliances. At present, the U.S. deployment in the Philippines “has effectively become equivalent to maintaining a permanent military presence” in the country.

III. Future Trends and Influencing Factors

For a long time, the U.S. and the Philippines have continuously deepened and broadened their bilateral alliance underpinned by agreements including MBA, MDT and EDCA. The fundamental drivers of progress in U.S.-Philippines relations are U.S. strategic competition with China and the Philippines’ strategy of deterring China. The U.S. exploits the Philippines to stir tension in the South China Sea and the Taiwan Strait, while the Philippines seeks to leverage U.S. power and capitalize on its geographic advantages to strengthen its own capabilities, conduct infringements and provocations in the South China Sea, and expand its international influence. However, constrained by domestic politics in both countries and different national interests, U.S. military deployments in the Philippines tend to follow a “light-asset” approach, avoiding excessive investment.

IV.Summary

As a major U.S. ally in the Asia-Pacific and a South China Sea claimant, the Philippines has long been regarded by the U.S. as a “strategic pawn” for stirring tension in the South China Sea and even the Taiwan Strait. Since Marcos Jr. took office in 2022, he has comprehensively restarted and upgraded the military alliance with the U.S., closely aligning Philippine strategy with Washington and shifting the country from “balanced diplomacy” to the “front line of confrontation.” Through EDCA, the U.S. continues to expand its military footprint on Philippine bases and conducts realistic combat exercises involving autonomous weapons and unmanned platforms to enhance bilateral and multilateral interoperability.

However, Philippine bases offer limited utility in wartime. Therefore, the Philippines’ strategic positioning is primarily as a supplement to U.S. military bases in Japan, and U.S. investment in base expansion has fallen short of expectations. To enhance interoperability and combat readiness, bilateral and multilateral joint exercises and training between the U.S. and the Philippines are expected to continue on the rise, strengthening their distributed denial capabilities in the Indo-Pacific. At the same time, to address the region’s lack of forward ammunition and equipment production capacity, the U.S. may push for the establishment of joint ammunition manufacturing and storage facilities, as well as ship repair facilities, at Subic Bay, thereby enhancing the Philippines’ ability to provide logistical support to U.S. forces.

The U.S.-Philippines military cooperation has leapt from a traditional “defensive alliance” to a key component of the Indo-Pacific “integrated deterrence” framework. The EDCA base network has completed a strategic encirclement layout against China that “locks the Taiwan Strait from the north and controls the South China Sea from the south.” Continuous investment in infrastructure expansion is turning the Philippines into a forward combat hub.

Deploying assets such as F-35, NMESIS, HIMARS, and unmanned surface vessels, the U.S. has achieved a shift from “rotational access” to “quasi-permanent presence” and from “bilateral exercises” to “multilateral joint operations”. Through a tiered cooperation network of the “core four countries” (U.S.-Philippines-Japan-Australia) plus extended partners, the U.S. has further embedded the Philippines within its Indo-Pacific strategy and command chain.

However, these forces are primarily a peacetime presence and would be extremely vulnerable in a high-end conflict. Their existence serves more political, strategic, and diplomatic purposes—using a relatively small military footprint to disrupt the situation, shape the environment, and push developments in a direction favorable to the U.S. Judging from U.S. investment and troop deployment, these bases are mainly intended to function as backup and logistics nodes in times of crisis or conflict, serving as supplements to U.S. bases in Japan.

Since 2013, although the U.S. has made significant progress in expanding Philippine bases and increasing its military presence there, the importance of these developments should not be overstated. The current situation in both the South China Sea and the Taiwan Strait remains tense but has not yet erupted into open conflict (“hot but not exploding”). The U.S. is willing to exploit these Philippine bases, actively using them to shape the strategic environment, contain China, and drain Chinese resources, yet it is unwilling to invest too heavily—especially since the existing infrastructure in the Philippines is poor and would require massive additional investment to achieve wartime resilience. The Philippine expectation that U.S. military presence will deter China is fundamentally unrealistic.

For China and other stakeholders, it is essential to objectively assess the trajectory of the U.S.-Philippine military alliance development and its impact on their own interests. While taking seriously the wartime threat posed by the U.S. military presence and deployments in the Philippines, greater attention should be paid to the restraints and attrition these arrangements and related activities impose on China during peacetime.

The backdrop to this speech is that on June 15, 2026, a national symposium on Party-building work was convened in Beijing, at which “Xi Jinping Thought on Party Building” was formally put forward, with its core being the “Fourteen Commitments.” Going a step further, Xi Jinping Thought on Party Building is now placed alongside Xi Jinping Thought on the Economy, on the Rule of Law, on Culture, on Ecological Civilization, on Strengthening the Military, on Diplomacy, and the Overall National Security Outlook.

This indicates that “Party building” has been formally elevated to a branch of theory ranking alongside the economy, the rule of law, diplomacy, and military strength. The speech’s heavy emphasis on full and rigorous Party self-governance suggests that Party building and anti-corruption efforts will continue to be advanced as “an important political task for the whole Party in the current period and for some time to come.” The threefold characterization of the anti-corruption campaign as a “tough, protracted, and all-out battle” indicates that there is no phased conclusion to these efforts; rather, they are being institutionalized and made a permanent feature.

Share

Below is the full text I made with the help of AI, which is not an official ver:

Speech at the Ceremony Marking the 105th Anniversary of the Founding of the Communist Party of China

(July 1, 2026)

Xi Jinping

Comrades and friends,

Today, we hold this grand gathering to celebrate the 105th anniversary of the founding of the Communist Party of China, to look back on the glorious journey our Party has traveled, to look ahead to the bright prospects for the development of the cause of the Party and the people, and to rally the whole Party and the people of all ethnic groups across the country to march forward with full confidence toward the grand goals of building China into a great modern socialist country in all respects and realizing the great rejuvenation of the Chinese nation.

First, on behalf of the Party Central Committee, I extend festive greetings to all members of the Communist Party of China! I extend warm congratulations to the recipients of the “July 1 Medal,” and to the outstanding Party members, outstanding Party workers, and advanced primary-level Party organizations across the country who have been honored!

Comrades and friends!

One hundred and five years ago, amid the great awakening of the Chinese people and the Chinese nation, and through the close integration of Marxism-Leninism with the Chinese workers’ movement, the Communist Party of China was born. From that moment on, the Chinese people and the Chinese nation had a most reliable backbone, and China—beset by internal turmoil and external aggression, impoverished and weak—embarked on earth-shaking historic changes.

Over the past 105 years, our Party has always adhered to its founding mission of seeking happiness for the Chinese people and rejuvenation for the Chinese nation. It has profoundly grasped the general trends of world development, accurately understood the changes in the principal social contradictions across different historical periods, and united and led the people of all ethnic groups in tireless struggle. It has achieved the great successes of the new-democratic revolution, of socialist revolution and construction, of reform and opening up and socialist modernization, and of socialism with Chinese characteristics in the new era, writing the most magnificent epic in the several-thousand-year history of the Chinese nation.

One hundred and five years of tireless struggle have fundamentally changed the future and destiny of the Chinese people. Leading the people through magnificent and heroic struggle, our Party overthrew the three mountains of imperialism, feudalism, and bureaucrat-capitalism, established a New China in which the people are masters of the country, brought a definitive end to old China’s history as a semi-colonial, semi-feudal society, and achieved the historic transformation of the people’s lives from inadequate food and clothing to moderate prosperity in general, and then to moderate prosperity in all respects. Today, the Chinese people have taken their destiny firmly into their own hands and are striding confidently toward an even brighter future with self-confidence, self-reliance, and self-improvement.

One hundred and five years of tireless struggle have opened up the correct path for realizing the great rejuvenation of the Chinese nation. Through the great practice of revolution, construction, reform, and the new era, our Party has led the people through countless hardships to successfully open up and adhere to the path of socialism with Chinese characteristics. In just a few decades, we have completed a process of industrialization that took developed countries several centuries, creating the two miracles of rapid economic development and long-term social stability. Today, the great rejuvenation of the Chinese nation is unstoppable, presenting brighter prospects than ever before.

One hundred and five years of tireless struggle have demonstrated the powerful vitality of Marxism. Our Party has adhered to integrating the basic tenets of Marxism with China’s specific realities and with China’s fine traditional culture, continuously advancing the adaptation of Marxism to the Chinese context and the needs of the times. It has developed Mao Zedong Thought, Deng Xiaoping Theory, the Theory of Three Represents, the Scientific Outlook on Development, and the Thought on Socialism with Chinese Characteristics for a New Era, greatly enriching and developing Marxism. Today, the vigorous vitality and dynamism of the cause of socialism with Chinese characteristics have fully tested the scientific and truthful nature of Marxism, fully demonstrated its people-centered and practical nature, and fully manifested its openness and contemporary relevance.

One hundred and five years of tireless struggle have profoundly influenced the course of world history. Our Party has always stood on the right side of history and on the side of human progress, and through unremitting self-improvement and struggle it has profoundly changed the trends and configuration of world development. Today, the Party has led the people in advancing Chinese modernization, creating a new form of human advancement and broadening the paths for developing countries to achieve modernization. We have promoted the building of a community with a shared future for mankind, contributing Chinese wisdom, Chinese solutions, and Chinese strength to solving major problems facing humanity. Socialist China under the Party’s leadership is widely recognized as a builder of world peace, a contributor to global development, and a defender of the international order.

One hundred and five years of tireless struggle have forged a strong Communist Party of China. Our Party has kept firmly in mind the nature, purpose, and objectives of a Marxist political party, has vigorously carried forward the great founding spirit of the Party, has grown ever more resolute through hardship, and has become ever stronger through repeated trials. Today, our Party has developed into the world’s largest governing party with significant global influence, enjoying the wholehearted support of the people. It is the strong leadership core of the cause of socialism with Chinese characteristics, and is fully worthy of being called a great, glorious, and correct party.

Comrades and friends!

Over the past 105 years, all the achievements we have made are the result of generation after generation of Chinese Communists uniting and leading the Chinese people in tireless struggle. The Chinese Communists represented chiefly by Comrades Mao Zedong, Deng Xiaoping, Jiang Zemin, and Hu Jintao established immortal, historic feats for the great rejuvenation of the Chinese nation! We pay them the highest tribute!

At this moment, we deeply cherish the memory of Comrades Mao Zedong, Zhou Enlai, Liu Shaoqi, Zhu De, Deng Xiaoping, and Chen Yun, and other revolutionaries of the older generation, as well as Comrade Jiang Zemin. We deeply cherish the memory of the revolutionary martyrs and people of noble aspiration who advanced wave upon wave and sacrificed themselves for national independence, the liberation of the people, the prosperity of the country, and the happiness of the people. The great achievements they made for the country and the nation will forever be recorded in history! Their noble spirit will forever be etched in the hearts of hundreds of millions of people!

The people are the creators of history and the true heroes. Here, on behalf of the Party Central Committee, I pay the highest tribute to the people across the country and to people from all walks of life who are striving on every front! I extend sincere greetings to our compatriots in the Hong Kong Special Administrative Region, the Macao Special Administrative Region, and Taiwan, and to overseas Chinese! I express heartfelt thanks to all the peoples and friends of other countries who have lived in friendship with the Chinese people and who have cared about and supported China’s cause of revolution, construction, and reform!

Comrades and friends!

The reason the Communist Party of China has been able to continually create brilliance over its 105 years of struggle, and the reason history and the people chose the Communist Party of China, lies fundamentally in the outstanding qualities our Party possesses that are incomparable to those of any other political party or political force.

— Devoted to the pursuit of truth, always keeping to the correct direction of advance. Our Party takes Marxism as a powerful ideological weapon for transforming both the subjective and objective worlds. It upholds emancipating the mind, seeking truth from facts, keeping pace with the times, and being realistic and pragmatic. It focuses on revealing and applying truth in the movement of social contradictions, is adept at testing and developing truth in practice, and continually guides new practice with new theory, ensuring the cause of the Party and the people advances in the correct direction.

— Deeply rooted in the people, always possessing a solid foundation. Our Party keeps firmly in mind that this country is its people and the people are this country. It upholds building the Party for the public good and exercising governance for the people, and consciously practices the fundamental purpose of serving the people wholeheartedly. It stands firmly together with the people, thinks together with the people, and works together with the people, possessing a solid foundation that no storm can shake.

— Courageously shouldering historical missions, always retaining the strategic initiative. Cherishing the lofty ideal of communism and dedicated to the everlasting cause of the Chinese nation, our Party firmly shoulders the heavy responsibilities entrusted to it by history and the people. It upholds the unity of long-term goals and stage-specific goals, establishes its central tasks in accordance with changes in the principal social contradiction, and formulates and implements correct lines, principles, and policies, ensuring it firmly holds the leadership and initiative in the development of the cause.

— Conforming to the trends of development, always standing at the forefront of the times. With a high degree of historical consciousness and a broad global vision, our Party has a clear grasp of China’s national conditions and the theme of the times. It actively recognizes, responds to, and seeks change; it forges ahead with determination; and it ensures the Party’s various endeavors reflect the character of the times, grasp underlying laws, and are full of creativity, leading the country and nation to advance within the logic of historical progress and to develop within the current of the times, becoming a well-deserved vanguard of the times and the backbone of the nation.

— Daring and adept at struggle, always maintaining confidence in victory. Our Party upholds and carries forward the spirit of fearing no sacrifice and fighting heroically. For the sake of the people, the country, and the nation, and for the sake of ideals and convictions, it cuts through thorns and forges ahead. No matter how powerful the enemy, how arduous the road, or how severe the challenge, it is always fearless and never retreats, inspiring the people across the country to move continually from victory to victory with an iron will that stands firm no matter how the storms may rage.

— Committed to strengthening itself, always full of vigor and vitality. Our Party is soberly aware that to forge iron, one must be strong oneself. It is adept at examining itself against the requirements of the development of the times, alerting itself with a strong sense of anxiety, and improving itself with the spirit of self-reform. It attaches great importance to its own development, resolutely eliminating all factors that harm the Party’s advanced nature and purity and clearing away all viruses that erode the Party’s healthy body, becoming stronger and more powerful through revolutionary tempering.

These outstanding qualities are the key to why the Communist Party of China is able to succeed. In light of new realities, we must continually carry forward and develop the Party’s outstanding qualities, ensuring the Party never changes its nature, its color, or its flavor, and that it always possesses strong creativity, cohesion, and fighting capacity.

Comrades and friends!

“Great accomplishments depend on ambition; great undertakings depend on diligence.” Fulfilling the Party’s mission and tasks on the new journey of the new era requires all Chinese Communists to firm up their confidence and continue striving, continually creating new achievements worthy of the times and the people.

— To firm up confidence and continue striving, we must uphold the Party’s basic theory, basic line, and basic policy. The Party’s basic theory, basic line, and basic policy are major achievements attained by the Party and the people through arduous exploration, and they are the fundamental guide for the cause of the Party and the country. On the new journey, the whole Party must maintain the resolve to keep to our course and never waver in our aspirations. We must uphold the Party’s overall leadership and the centralized, unified leadership of the Party Central Committee, thoroughly implement the Thought on Socialism with Chinese Characteristics for a New Era, stay firm in our confidence in the path, theory, system, and culture of socialism with Chinese characteristics, carry forward our cause into the future while upholding fundamental principles and breaking new ground, and ensure that we are not blinded by passing clouds and never lose our bearings as we ride the wind and break the waves.

— To firm up confidence and continue striving, we must rely closely on the people to create great historical undertakings. Upholding and developing socialism with Chinese characteristics is the common cause of the whole Party and the people of all ethnic groups across the country. On the new journey, the whole Party must further boost its drive and spirit for hard work and enterprise, uniting and leading the people of all ethnic groups to blaze new trails. We must advance the Five-Sphere Integrated Plan in a coordinated manner and the Four-Pronged Comprehensive Strategy, fully and faithfully apply the new development philosophy on all fronts, accelerate the creation of a new development pattern, and make solid progress in high-quality development. We must act on the people-centered philosophy of development, develop whole-process people’s democracy, continually consolidate and develop the great unity of all ethnic groups, of all the people across the country, and of all the sons and daughters of the Chinese nation, fully arouse the initiative, enthusiasm, and creativity of the hundreds of millions of people, and pool the mighty strength to advance Chinese modernization.

— To firm up confidence and continue striving, we must actively respond to the risks and challenges on the road ahead. China’s development is at a stage in which strategic opportunities coexist with risks and challenges, and uncertainties and unforeseeable factors are increasing; we must always be prepared to withstand major tests amid high winds and choppy waters, and even perilous storms. On the new journey, the whole Party must strengthen its sense of anxiety, adhere to bottom-line thinking, carry forward the fighting spirit, and enhance its ability to fight. We must better coordinate the two overall situations of the domestic and the international, coordinate development and security, and enhance our ability to scientifically foresee changes, promptly detect risks, and effectively respond to challenges, ensuring the great ship of China’s rejuvenation cleaves through the waves and sails steadily far into the future.

— To firm up confidence and continue striving, we must persistently promote the building of a community with a shared future for mankind. As the once-in-a-century changes accelerate, the world has entered a new period of turbulence and transformation, and humanity once again stands at a crossroads. On the new journey, we must go with the aspirations of the people and the trend of the times, hold high the banner of peace, development, cooperation, and mutual benefit, promote humanity’s common values, promote the building of a new type of international relations, and advance the implementation of the Global Development Initiative, the Global Security Initiative, the Global Civilization Initiative, and the Global Governance Initiative, injecting more positive energy into world peace and development.

— To firm up confidence and continue striving, we must persist unremittingly in advancing full and rigorous self-governance of the Party. Full and rigorous self-governance is always a work in progress. On the new journey, we must fully implement the thought on Party building for the new era and act on the general requirements for Party building in the new era. Focusing on enhancing the Party’s ability to govern over the long term, maintaining the Party’s advanced nature and purity, and maintaining the Party’s close ties with the people, we must improve the system for exercising full and rigorous Party self-governance, strengthen all aspects of Party building with political building as the guide, and resolutely wage the tough, protracted, and all-out battle against corruption. We must continually enhance the Party’s capacity for political leadership, ideological guidance, mobilization of the people, and appeal to society, ensuring the Party always remains the strong leadership core throughout the historical process of upholding and developing socialism with Chinese characteristics in the new era.

Comrades and friends!

To make the country strong, we must make the military strong, and only a strong military can ensure national security. On the new journey, we must fully implement the Party’s thought on strengthening the military for the new era, implement the military strategy for the new era, uphold the Party’s absolute leadership over the people’s armed forces, and stay firm on the path of building a strong military with Chinese characteristics. We must comprehensively advance the political building of the military, the reform of the military, the strengthening of the military through science and technology and through personnel, and running the military in accordance with law. We must advance the modernization of national defense and the armed forces in a high-quality manner, achieve the centenary goal of the People’s Liberation Army on schedule, and move faster to build the people’s armed forces into world-class forces, resolutely safeguarding China’s sovereignty, security, and development interests and making greater contributions to safeguarding world peace and development.

Promoting long-term prosperity and stability in Hong Kong and Macao is an inherent requirement of the great rejuvenation of the Chinese nation. On the new journey, we must fully, faithfully, and resolutely implement the policy of “One Country, Two Systems,” under which the people of Hong Kong govern Hong Kong and the people of Macao govern Macao, both with a high degree of autonomy. We must implement the principle of patriots administering Hong Kong and patriots administering Macao, enhance the effectiveness of law-based governance in Hong Kong and Macao, promote their economic and social development, and support them in better integrating into and serving the overall development of the country.

Resolving the Taiwan question and realizing the complete reunification of the motherland is a historic mission to which our Party is unswervingly committed and the common aspiration of all the sons and daughters of the Chinese nation. We must thoroughly implement the Party’s overall policy for resolving the Taiwan question in the new era, uphold the one-China principle and the 1992 Consensus, unite the broad masses of our Taiwan compatriots, deepen cross-Strait exchanges, cooperation, and integrated development, resolutely combat “Taiwan independence” separatist forces, oppose interference by external forces, and firmly advance the cause of national reunification.

Comrades and friends!

Young people are a vital force for realizing the great rejuvenation of the Chinese nation. The whole Party must value, care for, and support young people, and create the conditions for their growth and success. The young people of China in the new era must unswervingly heed the Party’s call and follow the Party, set lofty aspirations, courageously shoulder the responsibilities of the era, integrate their personal pursuits into the cause of the Party and the country, seize the day and make the most of their youth, run their leg well in the historic relay on the new journey, and pave the way with their youth while letting their ideals shine!

Comrades and friends!

The 105-year glorious history of the Communist Party of China is a source of pride, but we must never become arrogant or complacent, nor halt our progress. By the middle of this century, we are to build China into a great modern socialist country in all respects and achieve the Second Centenary Goal. Time waits for no one! History waits for no one! All comrades in the Party must stay true to our founding mission and keep it firmly in mind; we must remain modest and prudent and work hard; and we must dare to struggle and be adept at struggle. Relying closely on and uniting and leading the people of all ethnic groups across the country, let us forge ahead on the new journey and make new contributions in the new era, striving to create new historic glory!

More binding targets for new energy development

In the 14th Five-Year Plan for a Modern Energy System, there were no binding indicators governing new energy development; and in the 14th Five-Year Plan for Renewable Energy Development, there were only five anticipatory (non-binding) indicators concerning the development and utilization of renewable energy—namely, the total renewable electricity consumption responsibility weighting, the non-hydro renewable electricity consumption responsibility weighting, renewable power generation, total renewable energy utilization, and non-electric renewable energy utilization. The 15th Five-Year Plan for Building a New-Type Energy System, by contrast, adds three binding indicators: comprehensive energy production capacity, raising the share of non-fossil energy in total consumption to 25% by 2030, and reducing carbon emissions per unit of power generation by more than 10%.

The shift in wording from “modern energy system” to “new-type energy system” reflects a change in policy thinking

In the 15th Five-Year Plan, the phrasing is “accelerating the construction of a new-type energy system,”加快建设新型能源体系 whereas in the text of the 14th Five-Year Plan the formulation was “accelerating the building of a modern energy system and promoting high-quality energy development.”加快构建现代能源体系、推动能源高质量发展 It is worth noting that the concept of a “new-type energy system” first appeared in the report of the 20th National Congress of the Communist Party of China in October 2022, in the section on promoting green development:

Advance the energy revolution in greater depth, promote the clean and efficient use of coal, intensify the exploration, development, and expansion of oil and gas reserves and production, accelerate the planning and construction of a new-type energy system, coordinate hydropower development with ecological protection, develop nuclear power in an active, safe, and orderly manner, and strengthen the systems for energy production, supply, storage, and marketing to ensure energy security.

深入推进能源革命，加强煤炭清洁高效利用，加大油气资源勘探开发和增储上产力度，加快规划建设新型能源体系，统筹水电开发和生态保护，积极安全有序发展核电，加强能源产供储销体系建设，确保能源安全。

The related concept of a “new-type power system,” meanwhile, was put forward even earlier, at the Ninth Meeting of the Central Financial and Economic Affairs Commission in March 2021:

We must build a clean, low-carbon, safe, and efficient energy system, control the total volume of fossil energy, work to improve utilization efficiency, implement a renewable energy substitution initiative, deepen reform of the electricity system, and build a new-type power system with new energy as the mainstay.

要构建清洁低碳安全高效的能源体系，控制化石能源总量，着力提高利用效能，实施可再生能源替代行动，深化电力体制改革，构建以新能源为主体的新型电力系统。

To see why the shift happened, it helps to look at how each plan frames the energy system in the first place. Early in the 14th Five-Year Plan—Chapter 1, Article 2—the text surveys the global energy landscape and then declares that China’s energy security has entered a critical phase. Tellingly, the examples it reaches for first are all from the traditional side of the ledger, with oil and gas front and center; only afterward does it mention that the low-carbon transition has entered an important window. In Chinese policy documents, the order in which things appear reflects a certain hierarchy of priorities. Read that way, the plan’s real emphasis was still on securing energy through diversified oil and gas imports, building up petroleum reserves, and the like.

By the time we reach the 15th Five-Year Plan for Building a New-Type Energy System, however, the very first of its main objectives, stated at the outset, declares that “by 2030, a clean, low-carbon, safe, and efficient new-type energy system will be initially established.” It then goes on to state:

Coal and oil consumption will peak; the share of non-fossil energy in total consumption will reach 25%; wind and solar will account for more than 50% of installed power capacity, becoming the mainstay of installed capacity; and non-fossil energy will reach 50% of power generation, becoming the mainstay of electricity output.

煤炭和石油消费达峰，非化石能源消费比重达到 25%，风电和太阳能发电装机比重超过50%、成为电力装机主体，非化石能源发电量比重达到 50%、成为电量主体

As we can see, the change in the order of these formulations reflects how the focus of policy on safeguarding energy security has already shifted toward relying on the non-fossil energy domain. If that seems like reading too much into word order, the project lists make the same point more concretely. The 14th Five-Year Plan’s energy-security priorities were domestic oil and gas exploration, gas storage, LNG import terminals, and grid security. The 15th Five-Year Plan’s land on oil and gas reserves, pipeline buildout, and “high-standard” coal supply bases. In other words, the fossil-fuel measures have visibly contracted into a storage-and-backstop role.

It is worth noting that at the press conference, when Wan Jinsong万劲松, Deputy Director of the National Energy Administration, responded to a question about increased investment in coal-to-chemicals, he affirmed the role of coal in regulating power generation and maintaining electricity stability, yet placed particular emphasis on its positioning as “small in scale but stable.” Given that coal-fired power had long been China’s single largest source of electricity, this statement, in a sense, demotes coal from its status as a mainstay energy source to a supporting role for new energy—and it likewise reflects how China’s thinking on safeguarding energy security has shifted from a reliance on diversified imports of traditional fossil fuels toward a reliance on its technological advantages in the new energy domain.

From “West-to-East Power Transmission” and “West-to-East Gas Transmission” to “Western Energy for Western Use”

At the press conference, Wan Jinsong put forward a new formulation ”Western Energy for Western Use.” (西能西用) It offers a geographic dimension for understanding the new-type energy system. China’s resources are relatively unevenly distributed across regions, with coal, oil, and gas resources concentrated mainly in the west, while energy consumption comes chiefly from the industrial zones along the eastern seaboard. The government’s previous approach centered on one-directional flows that ship resources from west to east, such as West-to-East Gas Transmission (西气东输), West-to-East Power Transmission (西电东输), and North-to-South Coal Transport(北煤南运).

But as AI drives ever-growing demand for electricity, it has made computing power a uniquely suited resource for western regions, allowing electricity generated by wind, solar, and other new energy sources to be converted into computing power locally.

By Wan Jinsong’s account, the west will henceforth no longer simply export primary energy such as electricity, coal, and natural gas; instead, it will “guide energy-intensive industries to relocate to the west in an orderly fashion, and promote the coordinated layout of advanced manufacturing, computing power, hydrogen energy, and other industries alongside the development of clean energy bases,” converting resources into products and tokens on the spot, and keeping the added value local.

The 15th Five Year Plan For The Development Of A New Type Energy System（chinese）

473KB ∙ PDF file

Download

Download

English version I made with the help of AI:
https://rattle-garment-31c.notion.site/The-15th-Five-Year-Plan-for-the-Development-of-a-New-Type-Energy-System-38f41f93c20f80ce94dff81b0e2ade17?source=copy_link

Below is the transcript of the press release I made with the help of AI

Share

The State Council Information Office Holds a Press Conference on the Planning for Accelerating the Development of a New-Type Energy System during the 15th Five-Year Plan Period

Jia Huili, Deputy Director-General of the Press Bureau and Spokesperson, State Council Information Office (SCIO):

Ladies and gentlemen, good morning. Welcome to this press conference held by the State Council Information Office. Today we continue our series of themed press conferences under “Getting Off to a Good Start in the 15th Five-Year Plan.” We have invited Mr. Wang Hongzhi, member of the Leading Party Members Group of the National Development and Reform Commission (NDRC) and Director of the National Energy Administration (NEA), to brief you on the planning related to accelerating the development of a new-type energy system during the 15th Five-Year Plan period, and to answer your questions. Also attending today’s conference are Mr. Wan Jinsong, NEA Spokesperson and Deputy Director; Mr. Ren Yuzhi, NEA Spokesperson and Director-General of the Development Planning Department; and Mr. Du Zhongming, Director-General of the Electric Power Department.

Now, let us first invite Mr. Wang Hongzhi to give his briefing.

Wang Hongzhi, Member of the Leading Party Members Group of the NDRC and Director of the NEA:

Friends from the media, ladies and gentlemen, good morning! Thank you for your long-standing care and support for energy work. With the approval of the State Council, the Plan for Building a New-Type Energy System during the 15th Five-Year Plan Period (hereinafter “the Plan”) was recently released.

The overall goal of the Plan is that, by 2030, a clean, low-carbon, safe, and efficient new-type energy system will be initially established, providing strong support for building China into an energy power. After five years of development, China’s energy system will exhibit four characteristics:

First, greater security and resilience. Comprehensive energy production capacity will reach 5.8 billion tons of standard coal equivalent, with self-sufficiency capacity holding steady with a slight upward trend, and energy imports becoming more diversified and controllable.

Second, a more optimized energy mix. At present, China’s installed power generation capacity has already surpassed 4 billion kilowatts, and is expected to reach 5.4 billion kilowatts by 2030. Of this, new energy will account for more than 50% of installed capacity, becoming the main body of installed power capacity; non-fossil energy will account for 50% of power generation, becoming the main source of electricity; and coal and oil consumption will peak.

Third, a renewed system configuration. We will accelerate the construction of a new-type energy infrastructure system that is strong and resilient, green and low-carbon, integrated and converged, and intelligent and efficient, and initially build a new-type power system.

Fourth, stronger innovation vitality. The modernization level of the industrial chain will be significantly enhanced, placing China among the world’s leading nations in energy science and technology innovation; the market and pricing mechanisms suited to the new-type energy system will be more robust, and a unified national electricity market will be essentially established.

To achieve these goals, our task arrangements emphasize “three strengthenings.”

1. Strengthening the coordinated optimization of spatial layout. Adhering to the principle of “one chessboard for the whole country,” we will coordinate and optimize the layout and flows of energy. The focus is: targeting the large-scale development of non-fossil energy by building “four bases + distributed” five major growth sectors to support raising the share of non-fossil energy consumption to 25%; targeting the consolidation of the foundation for fossil energy production by clearly designating the Ordos Basin and other areas as strategic oil and gas security bases, and optimizing the development of five major coal production bases; and targeting the optimization of trunk corridor layouts by strengthening the construction of energy transmission “arteries” such as West-to-East Power Transmission and West-to-East Gas Transmission.

2. Strengthening strategic coordination in system building. Building a new-type energy system is a systematic undertaking, for which we have planned “six major systems” as strategic tasks. We will build a strong and resilient energy security guarantee system, enhancing capacities for strategic energy security, transition security, and emergency security. We will build an advanced and well-adapted new-type energy infrastructure system to support energy transition and development. We will build a green, low-carbon energy consumption system, advancing energy conservation and carbon reduction in key sectors and deeply implementing clean energy substitution. At the same time, we will build a self-reliant and self-strengthening energy science and technology innovation system, a coordinated and efficient modern energy governance system, and a multidimensional and diversified energy international cooperation system.

3. Strengthening support driven by engineering projects. On one hand, we will implement a batch of major strategic projects. We will advance high-quality development of major projects such as the Lower Yarlung (Yaxia) Hydropower Project, build new energy bases in deserts, the Gobi, and arid regions (”shage huang”), water-wind-solar integrated bases, and offshore wind power bases, and actively, safely, and orderly advance coastal nuclear power construction. We will coordinate the construction of infrastructure such as the “national oil and gas network” and new-type power grids. On the other hand, we will actively develop new-industry and new-business-model projects. We will upgrade the quality of the electric vehicle charging infrastructure network, lay out and build wind-solar-hydrogen-ammonia-methanol integrated bases, build concentrated solar power (CSP) projects, advance direct green power connection and smart microgrid development, and complete a batch of zero-carbon industrial parks. It is projected that investment in key energy projects and new business models during the 15th Five-Year Plan period will maintain steady growth, with a total scale exceeding 20 trillion yuan.

In response to the internal and external situation of energy development, and centered on security guarantees, low-carbon transition, and innovation-driven development, we have also planned “three priorities.”

First, prioritizing the strengthening of oil and gas security. We will strengthen domestic production and vigorously advance oil and gas exploration and development. We will strengthen international cooperation and consolidate and expand diversified oil and gas imports. We will increase efforts in new energy development, strengthen the strategic deployment of various alternative fuels, and advance the large-scale application of new-energy heavy-duty trucks and ships.

Second, prioritizing the construction of the new-type power grid. The new-type power grid is one of the “six networks” that society pays close attention to. We will optimize the layout of inter-provincial and inter-regional transmission corridors, strengthen coordinated construction and operation of grids at all levels, and build a new-type power grid that is safe and reliable, green and low-carbon, strong and resilient, and intelligent and flexible—coordinating the main grid, distribution grid, and microgrids.

Third, prioritizing the mutual empowerment of AI and energy. We will deeply implement the “AI+” energy initiative and accelerate the digital and intelligent transformation of the energy system. At the same time, we will strengthen the coordinated development of computing and power, coordinating energy resource allocation with computing infrastructure construction, promoting “stronger computing through power, and better power through computing.”

Friends from the media, ladies and gentlemen, the blueprint is set, and the future is promising! I am confident that, after five years of sustained effort, a new-type energy system with greater security and resilience, a more optimized energy mix, a renewed system configuration, and stronger innovation vitality will take shape more quickly, laying a solid foundation for building China into an energy power!

That concludes my briefing. My colleagues and I will now continue to exchange views with you in detail. Thank you!

Reporter, Hong Kong Ta Kung Wen Wei Media Group:

Currently, international geopolitical conflicts continue to disrupt oil and gas supply chains, while the rapidly rising share of wind and solar power domestically brings risks of system fluctuations. During the 15th Five-Year Plan period, what targeted measures will China take to further strengthen its energy security capacity and comprehensively enhance energy self-sufficiency and system risk resilience? Thank you.

Wang Hongzhi:

Thank you for your question. I will answer this one.

Energy supply and security bear on the national economy and people’s livelihoods. Since the 18th Party Congress, General Secretary Xi Jinping has put forward the new energy security strategy of “four revolutions and one cooperation,” emphasizing that we must firmly hold the energy rice bowl in our own hands. In recent years, under the strong leadership of the Party Central Committee and the State Council, the NDRC and NEA have deeply implemented the new energy security strategy, working simultaneously on energy security and green, low-carbon transition, and effectively responding to various domestic and international risks and challenges. In particular, since late February this year, the global energy market has experienced severe turbulence, with many countries experiencing varying degrees of energy shortages, surging oil prices, and tight supplies. Yet China’s energy system withstood the shock, maintaining overall balance between supply and demand and overall price stability, demonstrating tremendous resilience. International media commented that China not only successfully resisted the external shock but also made important contributions to stabilizing the global energy market.

The 15th Five-Year Plan period is a critical phase for laying a solid foundation and making full efforts toward basically achieving socialist modernization, during which energy demand will continue to grow. We will always place energy security in the foremost position, adhere to bottom-line thinking and systems thinking, and focus on three areas of work.

First, doing well in the “development” and “reserves” of fossil energy. Coal is our greatest source of confidence for stable energy supply, and we must build coal bases based on this national condition. For oil and gas, we will increase domestic exploration and development efforts, stabilize oil output, and increase natural gas supply. At the same time, we will strengthen energy reserves, improve the oil and gas reserve system, strengthen coal-to-oil/gas production capacity and technology reserves, and improve coal production capacity reserves—only then can we remain “as steady as a mountain through wind and rain.”

Second, significantly increasing the supply of non-fossil energy. This is both a requirement of the transition and a way to expand new space for energy security. During the 14th Five-Year Plan period, 45% of the increment in China’s energy consumption was already met by non-fossil energy, and this proportion will rise further during the 15th Five-Year Plan period. We will adhere to the simultaneous development of multiple energy sources such as wind, solar, hydro, and nuclear, build a new-type power system, promote large-scale and high-proportion development and utilization of new energy, and accelerate the supply of newly added electricity consumption from clean energy.

Third, expanding the circle of energy cooperation partners. As everyone knows, a large proportion of China’s oil and gas currently comes from the international market. We will adhere to the diversification of energy imports, strengthen mutually beneficial cooperation with key energy and resource countries, work with other nations to safeguard energy transport channels, and ensure energy security under open conditions.

The Plan makes specific arrangements for each of these priorities, with the aim of building a strong and resilient energy security guarantee system.

Reporter, Upstream News:

The 15th Five-Year Plan period is the decisive period for achieving the carbon peaking goal, and the energy sector is the main battlefield for carbon peaking. What indicators has the Plan for Building a New-Type Energy System during the 15th Five-Year Plan Period set in areas such as raising the share of non-fossil energy supply, advancing the orderly substitution of fossil energy, and energy conservation and carbon reduction? What specific measures will be taken to ensure the carbon peaking goal is met on schedule? Thank you.

Ren Yuzhi, NEA Spokesperson and Director-General of the Development Planning Department:

Thank you for your question. Achieving the carbon peaking goal on schedule is China’s commitment to the international community. Currently, the energy sector accounts for more than 80% of carbon dioxide emissions—indeed the main battlefield for emissions reduction that you mentioned. During the 15th Five-Year Plan period, we will accelerate the low-carbon transition with greater intensity and more concrete measures, making green the defining backdrop of high-quality development.

In terms of development goals, we have set five binding indicators around transition and decarbonization. Three of these—the share of non-fossil energy consumption, the share of non-fossil energy in power generation, and the share of electricity in final energy consumption—are carried over from the 14th Five-Year Plan, and we have set higher targets for them. At the same time, we have established two new indicators. One is the carbon emissions per unit of electricity generated, with a 15th Five-Year Plan target of reducing it by more than 10%; this is a comprehensive indicator reflecting the green transition and efficiency improvement of China’s power system. The other is the energy savings in key industries; through energy conservation and carbon reduction retrofits and technological innovation, key industries will save more than 150 million tons of standard coal equivalent during the 15th Five-Year Plan period.

We have the roadmap; the key is implementation. In terms of specific measures, we will exert efforts in the same direction on both the energy supply side and the consumption side to reduce carbon collaboratively. On the supply side, we will work on both increment and stock. Regarding increment, the main task is to scale up clean energy supply. We will vigorously promote large-scale development of non-fossil energy; by the end of the 15th Five-Year Plan period, installed non-fossil energy capacity is expected to exceed 3.5 billion kilowatts, an increase of nearly 50% over the end of the 14th Five-Year Plan period. Of this, wind and solar power capacity will reach more than 2.8 billion kilowatts, laying a solid foundation for achieving the 2035 nationally determined contribution (NDC) targets. Regarding stock, the main task is the transition of traditional energy. We will strengthen green coal mining and clean production and utilization, promote green, low-carbon development of oil and gas extraction and refining, strengthen the integration of coal, oil and gas, and coal power with new energy, and implement a new generation of coal power retrofits and upgrades.

On the consumption side, we will focus on two aspects: conservation and substitution. First, doing well in energy conservation and efficient utilization. We will strengthen the application of energy-saving and efficient production processes and technologies in the industrial sector, promote energy conservation and carbon reduction retrofits in traditional industries as well as energy efficiency improvements in new high-energy-consuming industries such as computing power, strengthen building energy efficiency management, and deepen energy conservation and carbon reduction retrofits in public institutions. Here, I also ask our friends in the media to help promote energy conservation—every kilowatt-hour of electricity and every drop of oil comes hard-won. Second, doing well in clean substitution of energy use. We will implement a system of minimum-share targets for renewable energy consumption, accelerate the green and low-carbon transition of heating systems, promote the establishment of around 100 national-level zero-carbon industrial parks, advance building-integrated photovoltaics (BIPV), promote the large-scale application of new-energy heavy-duty trucks, and build low-carbon and zero-carbon transport corridors. At the same time, we will cultivate and expand the green certificate and green power markets, strengthen the accounting of non-fossil energy power consumption, accelerate the improvement of policy mechanisms, unleash the energy conservation and carbon reduction potential in production and daily life, and continuously raise the level of green, low-carbon energy consumption. Thank you!

Reporter, China Economic Information Service (Xinhua):

During the 14th Five-Year Plan period, private enterprises made important contributions in the energy sector. During the 15th Five-Year Plan period, which specific areas will we focus on, and what measures will we take to further expand private investment and stimulate the vitality of private enterprises? Thank you.

Wan Jinsong, NEA Spokesperson and Deputy Director:

Thank you for your question. Private enterprises have strong innovation vitality and have always been a vital force in promoting high-quality energy development. At present, China’s green, low-carbon energy transition is accelerating, and new models and new business forms are emerging rapidly, providing a new “blue ocean” for the development of private enterprises. During the 15th Five-Year Plan period, we will adhere to open development and service-oriented investment promotion, and intensify support for private enterprises participating in building the new-type energy system.

In major energy projects, we will expand the investment space for private enterprises. For major projects with certain returns—such as nuclear power, hydropower, and oil and gas storage and transport facilities—we will assess the feasibility of private enterprise participation case by case. Currently, private enterprises have participated in 12 nuclear power projects, with the highest equity stake reaching 20%. During the 15th Five-Year Plan period, we will continue to open the door wide for energy projects, issue guidelines for private enterprise participation in investment in large and medium-sized hydropower and other projects, so that enterprises have direction for their investments and assurance of returns.

In new models and new business forms, we will support private enterprises in fully demonstrating their capabilities. Among the 105 direct green power connection projects already approved, private enterprise projects account for half. The multi-user direct green power connection policy has now been issued, further stimulating the enthusiasm of private enterprises to invest in zero-carbon parks, industrial green microgrids, and integrated energy services. We will further improve the electricity market and pricing mechanisms, and support private enterprises in investing in virtual power plants, charging facilities, new-type energy storage, and other projects.

In optimizing the business environment, we will provide better services for private enterprises. We will concentrate on rectifying chaos in bidding and tendering, disorderly competition in the photovoltaic industry, and compulsory industrial bundling requirements, and continuously advance the fair opening of infrastructure such as power grids and oil and gas pipeline networks to private enterprises. We will simplify project approval processes, implement “one-network handling, one-window acceptance,” improve the level of “access to electricity” services, and make it easier for enterprises to get things done with fewer trips. We will regularly conduct government-enterprise dialogues, evaluate policy effects, and respond to development demands, serving as a trusted partner of enterprises.

During the 15th Five-Year Plan period, we will also successively introduce new reform and innovation measures, giving private enterprises greater investment space, more opportunities, and stronger confidence in the energy sector. Thank you!

Reporter, Haibao News:

In recent years, China’s total electricity consumption has repeatedly hit new highs. Through the joint efforts of all parties, energy and power supply has remained stable and orderly. Affected by factors such as climate change and the rising share of new energy capacity, how will electricity supply security be ensured during the 15th Five-Year Plan period? And what is the state of preparation for meeting peak energy demand this summer? Thank you.

Du Zhongming, Director-General of the Electric Power Department, NEA:

Thank you for your question. Last year, both China’s total electricity consumption and peak load hit new highs, and we adopted proactive and effective measures so that power supply remained stable and orderly throughout. During the 15th Five-Year Plan period, with the surge in computing-power electricity use, the rapidly increasing demand from EV charging facilities, and the rigid growth in electricity use from other industries and residential life, we project that the average annual increment in national electricity consumption will reach around 600 billion kilowatt-hours—equivalent to adding the electricity consumption of a medium-sized economy every year.

To this end, we will exert coordinated efforts across all links of generation, grid, load, and storage, adopting more forceful and effective measures. On the generation side, we will accelerate the formation of a green, low-carbon supply structure while continuing to leverage the backstop role of gas power and coal power. On the grid side, we will accelerate the construction of the new-type power grid to enhance power mutual-support capacity and security resilience. On the load side, we will vigorously develop virtual power plants to stimulate coordinated response capacity on the demand side. On the storage side, we will effectively leverage the role of new-type energy storage to enhance peak-shaving capacity during peak demand periods. In short, under the strong leadership of the Party Central Committee and the State Council, and with the strong coordination of relevant departments, the NDRC and NEA will make every effort to ensure supply, and we are confident and capable of guaranteeing national electricity supply during the 15th Five-Year Plan period.

The work of ensuring supply during this summer’s peak is now in full swing. Our assessment is that the national balance between electricity supply and demand will be generally maintained, though some provinces in the East China, Central China (including Southwest), and Southern regions face tight supply. Through means such as inter-provincial and inter-regional power mutual support, electricity supply in these areas can be effectively guaranteed. Going forward, we will conduct daily rolling assessments of the national electricity supply-demand situation, strengthen monitoring and early warning of hydropower station inflows and power plant coal and gas reserves, ensure that supply-guaranteeing power sources are put into operation wherever possible, consolidate the backstop role of coal power, and ensure that gas power can “step up and generate steadily” during peak periods. At the same time, we will leverage the role of the large grid, large corridors, and large market, fully utilizing the transmission capacity of transmission corridors and surplus resources across regions to conduct staggered, off-peak inter-provincial market-based mutual-support trading. For areas with tight supply, we will formulate “one province, one policy” work measures to respond promptly and effectively to potential supply shortages in individual provinces under extreme weather and other circumstances, resolutely holding the bottom line of large-grid security and people’s livelihood electricity use. Thank you!

Reporter, Bloomberg:

Since 2021, China has accelerated the approval of new coal-fired power plants. We have also seen a significant increase in investment in coal chemical projects. What are the drivers of this trend? Will this expansion continue to the end of this Five-Year Plan period? And how should the balance between energy security and climate goals be struck? Thank you.

Wan Jinsong:

Thank you for your question. Globally, the utilization of fossil energy has two universal characteristics. The first characteristic is that thermal power remains the cornerstone for ensuring the safe and stable operation of power systems in various countries. Countries like those in Europe and the US rely mainly on gas-fired power for support and regulation, while China, owing to its own energy endowment, relies mainly on coal power for regulation. Currently, China’s thermal power accounts for about 40% of installed capacity, which is below the world average. In recent years, China’s new energy development has achieved world-renowned results, contributing more than half of the world’s newly added wind and solar power capacity—and behind such rapid development is the flexible regulation capacity provided by thermal power. China has coordinated the transition of traditional energy with the rapid development of new energy, forging a path of high-quality development that is clean, low-carbon, safe, and efficient. Since the start of the 14th Five-Year Plan period, the share of coal power in installed capacity has dropped from 49% to 32%, and its share of power generation has dropped from 61% to 51%, while coal power’s pollutant emission standards are now on par with gas power. During the 15th Five-Year Plan period, to support the rapid development of new energy and ensure energy demand, coal power will continue to play its role.

The second characteristic is that global fossil energy utilization is gradually shifting from fuel to feedstock. In recent years, China has continuously strengthened the clean and efficient utilization of coal, improved the comprehensive utilization efficiency of coal as a chemical feedstock, promoted the high-end, diversified, and low-carbon development of the coal chemical industry, and accelerated the transformation and upgrading of the coal industry.

Regarding energy security and climate goals, I believe you have all seen that, since the 18th Party Congress, China has coordinated the advancement of energy security and low-carbon transition, building the world’s largest renewable energy system while also effectively safeguarding energy security. During the 15th Five-Year Plan period, we will continue to do this work well. First, for non-fossil energy, we will pursue “abundant and excellent”—that is, developing non-fossil energy with greater intensity, accelerating the enhancement of new energy’s safe and reliable substitution capacity, vigorously developing new regulating capabilities such as new-type energy storage, and reducing dependence on fossil energy as much as possible. Second, for fossil energy, we will pursue “minimal and stable”—that is, basing ourselves on safeguarding energy security and scientifically and rationally deploying the development and utilization of fossil energy according to the system’s minimum needs, so as to better support economic and social development with stable energy supply. Thank you!

Reporter, 21st Century Business Herald:

China’s artificial intelligence industry continues to grow in scale and is expected to exceed 10 trillion yuan by the end of the 15th Five-Year Plan period. What are the specific application scenarios of AI in the energy sector? Behind AI development is computing power, and behind computing power is electricity—how will China advance the coordinated layout of computing and power? Thank you.

Wang Hongzhi:

Thank you for your attention. I will answer this question. Currently, AI and energy have entered the fast lane of mutual empowerment. The application scenarios of AI in the energy sector are mainly reflected in three aspects.

In terms of improving quality and efficiency: AI assists in completing complex but repetitive labor such as intelligent inspection and defect identification, improving work efficiency. For example, the “Kunlun” large model in the petroleum industry has improved the computational efficiency of oil and gas exploration and development by more than tenfold, providing more precise and efficient decision-making support for optimizing oilfield development plans.

In terms of solving difficult problems: As high proportions of new energy are integrated on a large scale, power systems become more complex, and AI can effectively solve many problems that traditional methods struggle to address. For example, the “Yudian” (Power Control) large model in the power industry responds in real time to fluctuations in wind and solar output, assisting in generating grid dispatch plans, and can further improve the level of new energy accommodation.

In terms of intelligent operation: The application of AI is gradually leaping from assisted decision-making to autonomous execution. For example, when a fault occurs in the distribution grid, the “Guangming Electric Power” large model can act like a “doctor,” sensing line anomalies in real time, precisely locating fault points, and achieving intelligent repair.

At the same time, the explosive growth of AI has driven a surge in electricity consumption. For example, every 5 seconds of high-definition video generated by AI consumes electricity equivalent to fully charging 10 mobile phones. In response to this explosive growth, we will, in accordance with the requirement of “stronger computing through power, and better power through computing,” coordinate energy resource allocation with computing infrastructure construction and advance computing-power coordination across multiple dimensions.

First, strengthening coordination in planning and layout. In the western region, we will coordinate the layout of national computing hubs with the planning and construction of large new energy bases, and advance the coordinated construction of computing facilities and power systems. In the eastern region, we will promote coordinated planning of distributed computing with distributed power sources, microgrids, and virtual power plants, responding to computing demand locally.

Second, strengthening coordination in the policy system. We will encourage qualified computing facilities to adopt direct green power connections, support computing facilities in participating in green power and green certificate trading, encourage computing facilities to participate in the electricity market as flexible regulating resources, and promote positive interaction between computing facilities and the new-type power system.

Third, strengthening coordination in operational regulation. Based on the electricity-use characteristics and regulation potential of different computing tasks, we will optimize the arrangement of computing loads. For example, for computing tasks requiring high reliability and low latency, we will focus on ensuring stable power supply; for computing tasks with higher latency tolerance, we will guide them to proactively adapt to the grid’s peak-shaving needs. Thank you!

Reporter, China Electric Power News (China Energy Media Group):

The 15th Five-Year Plan proposes that, by 2030, a unified national electricity market will be essentially established. Currently, new energy is entering the grid on a large scale and in high proportions, posing new challenges to the electricity market’s pricing mechanisms and inter-regional mutual support. How will we improve market development going forward to better adapt to these new changes? Thank you!

Ren Yuzhi:

Thank you for your question, and thank you for your concern about the national electricity market. It can be said that, during the 14th Five-Year Plan period, the unified national electricity market system was initially established, achieving multi-tiered and multi-category coordination, with the functions of energy, ancillary services, green power and green certificates, and retail markets complementing each other. Coal power and commercial and industrial users have fully entered the market, and the market has become an important means of allocating electricity resources.

Regarding the “new challenge” you mentioned—that the rapid development of new energy poses to the electricity market—during the 15th Five-Year Plan period, we will anchor on the goal of essentially establishing a unified national electricity market, focusing our efforts on three aspects: “market channels, trading cycles, and governance capacity” to further build up the electricity market.

First, unblocking the market channels for accommodating new energy. Electricity requires real-time balancing, and the market also needs to respond dynamically. New energy generation is unstable, and the electricity market can strengthen regulation to accommodate it, promoting optimal resource allocation. For example, in intra-provincial electricity trading markets, we will push the spot market to fully enter formal operation and improve market categories such as ancillary services and retail, so that the potential of regulating power sources and demand-side resources is fully realized. In inter-provincial and inter-regional electricity trading, medium- and long-term trading provides the “fundamental base” for cross-provincial accommodation of new energy, while inter-provincial spot trading and regional inter-provincial mutual-support trading serve as “adapters” for coping with short-term wind and solar fluctuations. By conducting cross-grid-operating-area trading on a regular basis, we have also achieved real-time trading between the State Grid and China Southern Power Grid, enabling surplus wind and solar power to travel thousands of miles into millions of households, and making fuller and better use of transmission corridor capacity.

Second, optimizing trading cycles adapted to the new-type power grid. We will promote electricity market trading cycles to coordinate in two directions—”longer and shorter”—to more precisely match the generation characteristics of new energy. On the “longer” side, this means extending cycles and stabilizing expectations, encouraging both generators and users to sign multi-year electricity trading contracts and realizing year-round, continuous operation of the medium- and long-term market, so that the green value of new energy can always be seen and locked in. On the “shorter” side, this means shortening cycles and trading flexibly, making good use of short-term markets such as spot and ancillary services to precisely reflect the differing value of electricity in time and space, adapting to the high-frequency, flexible trading needs of new energy, new-type energy storage, virtual power plants, and others, thereby fitting the new-type power grid with a stable “regulator.”

Third, enhancing the governance level of the unified electricity market. Now that the market is built, it must be used well and remain stable. We will strengthen multi-departmental coordinated supervision, building a collaborative management system with multi-party linkage among government competent departments, energy regulatory agencies, trading institutions, and market management committees. We will adhere to measuring everything by one ruler, resolutely break down local protectionism and market barriers, resolutely rectify behaviors such as administrative interference in the market, price collusion, and abuse of market dominance, uphold unified market rules, closely monitor risks throughout the retail market process, and comprehensively improve market governance.

Reporter, The Cover:

The Fourth Plenary Session of the 20th CPC Central Committee made arrangements for “new-type energy infrastructure” for the first time, and the 15th Five-Year Plan proposes intensifying the construction of new-type energy infrastructure. What are the characteristics of new-type energy infrastructure? And how will this work be implemented and advanced?

Wan Jinsong:

Thank you for your question. As we understand it, new-type energy infrastructure is the “skeleton” and “hard support” of the new-type energy system—modern infrastructure that adapts to the energy revolution and serves transition and development. It enables clean energy such as wind, solar, and hydropower to be transmitted out, received, and used well, with distinct characteristics of being strong and resilient, green and low-carbon, integrated and converged, and intelligent and efficient. During the 15th Five-Year Plan period, we will focus on three directions to further accelerate the layout and construction of new-type energy infrastructure.

The first direction is accelerating the layout of energy infrastructure that promotes new energy development. We will optimize development methods, build wind and solar bases in the “Three North” regions and water-wind-solar integrated bases in the Southwest, develop direct green power connections according to local conditions, build low-carbon and zero-carbon oil and gas fields, and promote integrated and converged development of new energy. At the same time, we will enhance the integration and accommodation capacity of new energy, accelerate the building of a new-type power grid that is safe and reliable, green and low-carbon, strong and resilient, and intelligent and flexible, and promote growth of source and storage regulating capacity by more than 40%.

The second direction is accelerating the layout of energy infrastructure driven by new technologies and new models. Such infrastructure includes renewable hydrogen production, green fuels, charging facilities, comprehensive nuclear energy utilization, virtual power plants, integrated energy stations, and more. Here are a set of figures: by 2030, China’s charging infrastructure will reach 40 million units, achieving a doubling of development, and charging will become more convenient in the future. The scale of renewable hydrogen production will reach 2 million tons, used in steel and chemical production and other applications, making heavy industry cleaner. The regulating capacity of virtual power plants will reach more than 50 million kilowatts, equivalent to 50 large thermal power units that burn no coal.

The third direction is accelerating the digital and intelligent transformation of traditional energy infrastructure. Some time ago, we organized an “AI+” energy integration pilot, launching 51 high-value application scenarios in fields such as smart grids, smart coal mines, smart oil and gas fields, and smart power plants, comprehensively raising the intelligence level of infrastructure. Next, we will continue to deepen the opening of application scenarios and accelerate their implementation.

Reporter, National Business Daily:

The central authorities have deployed the coordinated advancement of the “six networks,” including the water network and the new-type power grid, with the new-type power grid being the core support of the new-type energy system. During the 15th Five-Year Plan period, the rapid growth of new energy capacity places higher demands on grid carrying and regulating capacity. What systematic arrangements does the NEA have for new-type power grid construction? And how will diverse electricity demands be guaranteed?

Du Zhongming:

Thank you for your question. Since the 18th Party Congress, China’s grid development has achieved world-renowned results. We have built the grid with the world’s largest scale of integrated new energy, as well as the large grid with the highest transmission voltage levels and the strongest transmission capacity. This large grid carries more than 10 trillion kilowatt-hours of electricity consumption annually while consistently maintaining safe and stable operation. Currently, the length of transmission lines of 220 kV and above nationwide exceeds 1 million kilometers, the scale of integrated new energy exceeds 1.8 billion kilowatts, 24 ultra-high-voltage direct-current arteries have been built, and the scale of West-to-East Power Transmission has reached 340 million kilowatts. More and more western green power is now being “flash-delivered” to eastern load centers through this large grid.

The new-type power grid you mentioned is an important component of the “six networks” major infrastructure laid out by the state, and it is also an important carrier for building the new-type energy system and new-type power system. Building the new-type power grid can more forcefully promote the development and accommodation of new energy, more flexibly support the high-speed development of computing power, more reliably ensure power supply quality for emerging industries, and more efficiently meet EV charging demand. At present, in accordance with the unified deployment of the Party Central Committee and the State Council, we are accelerating the formulation of an implementation plan for new-type power grid construction, accelerating the building of a new-type power grid that is safe and reliable, green and low-carbon, strong and resilient, and intelligent and flexible. During the 15th Five-Year Plan period, national grid fixed-asset investment will reach more than 5 trillion yuan. Going forward, we will focus on “three news.”

The first “new” is improving the grid’s new architecture. We will work to build a new-type power grid platform with coordinated development of the main grid, distribution grid, and microgrids, strengthen grid security and resilience, and hold firm the bottom line of safe large-grid operation. During the 15th Five-Year Plan period, we will newly commission 15 ultra-high-voltage DC green-power mega-corridors, raising the scale of West-to-East Power Transmission to over 420 million kilowatts. At the same time, we will build more than 6 mutual-support projects, raising power mutual-support capacity by about 40 million kilowatts, promote the digital and intelligent upgrading of distribution grids, accelerate smart microgrid development, upgrade charging networks, promote vehicle-grid interaction, accelerate the construction of high-power charging facilities and battery-swap facilities for electric heavy-duty trucks, and support and guarantee the charging needs of more than 110 million electric vehicles.

The second “new” is making breakthroughs in key new technologies. We will accelerate the innovative application of grid-forming technology, long-duration energy storage technology, and 100% new-energy mega-base transmission technology, and implement the “AI+ grid” project to equip the grid with an intelligent “brain.”

The third “new” is rolling out more and better new services. During the 15th Five-Year Plan period, we will continuously optimize new energy grid-connection and transmission services and substantially enhance the system’s regulating capacity. By 2030, the scale of new-type energy storage will reach 300 million kilowatts, serving and supporting the high-level accommodation of more than 2.8 billion kilowatts of new energy. We will strengthen computing-power coordination, serving and supporting the electricity needs of computing power with different characteristics, such as intelligent computing, general computing, and supercomputing. We will effectively improve power supply quality, advance the inclusive provision of power services, and accelerate the transition from “having access to electricity” to “using electricity well” and “using green electricity.” Thank you!

Reporter, Red Star News:

My question is: The Plan for Building a New-Type Energy System during the 15th Five-Year Plan Period proposes to coordinate “western power for western use” and “West-to-East Power Transmission.” How will China’s regional energy deployment be coordinated in the future? And what new changes will occur in the national pattern of energy production and consumption?

Wan Jinsong:

Thank you for your question. China’s energy production is mainly concentrated in the western region, while energy consumption is mainly concentrated in the eastern and central regions. Over the long term, an overall pattern of “West-to-East Power Transmission,” “West-to-East Gas Transmission,” “West-to-East Coal Transport,” and “North-to-South Coal Transport” has taken shape. During the 15th Five-Year Plan period, while expanding “western energy delivered eastward,” we will intensify the advancement of “western energy for western use.”

“Western energy delivered eastward” must further expand in scale to effectively ensure the energy needs of the eastern and central regions. In the power sector, we will accelerate the construction of clean energy bases such as wind and solar in the “Three North” regions, along with outbound transmission corridors, with an expected addition of more than 80 million kilowatts of West-to-East Power Transmission capacity. In the oil and gas sector, we will accelerate the construction of West-to-East Gas Transmission corridors. In the coal sector, we will continuously optimize coal transport layout and work to expand the scale of Xinjiang coal exports. At the same time, we will promote the tapping of potential in key energy-consuming areas in the central and eastern regions, strengthen the nearby development and utilization of local non-fossil energy, rationally lay out new supporting power sources, and promote “electricity from afar, and also from nearby.”

“Western energy for western use” must further strengthen capacity coordination and steadily guide the optimization of productivity layout. We will guide high-energy-consuming industries to transfer to the western region in an orderly manner, promote coordination between the layout of advanced manufacturing, computing power, hydrogen energy, and other industries and the development of clean energy bases, and promote the optimization of the spatial layout of energy-consuming industries. In the future, the western region will not only deliver electricity, coal, and gas outward, but also deliver products and tokens outward.

Going forward, as “western energy for western use” and “western energy delivered eastward” are advanced in depth, new changes will also emerge in the national pattern of energy production and consumption. Two changes are worth noting here. First, the trend of integrating clean energy with industry will accelerate, with more high-energy-consuming, low-emission industries and projects landing in the west and using clean energy; capacity integration will become a new driver of western economic development. Second, new forms of clean energy transmission will emerge. The Plan proposes deploying a batch of green hydrogen-ammonia-methanol projects in the western and northern regions, strengthening guidance on hydrogen pipeline network layout, so that cross-provincial and cross-regional outbound transmission of new energy will not only take the form of electricity but can also be delivered in the form of hydrogen energy and green fuels.

Reporter, CCTV (China Media Group):

It was just mentioned that investment in key energy projects and new business models during the 15th Five-Year Plan period will exceed 20 trillion yuan. What major engineering projects has the Plan for Building a New-Type Energy System during the 15th Five-Year Plan Period arranged? And what role will this investment play in driving economic growth?

Wang Hongzhi:

Thank you for your question. I will answer this final one.

In the 15th Five-Year energy plan, there are both “large” projects for energy production and transmission and “new” projects that embody new-quality productive forces in energy. Overall, these projects can be divided into three major categories:

The first category is for strengthening energy security. This mainly includes oil and gas, coal, supporting power sources, and coal-to-oil/gas, which form the fundamental base for energy supply security. Future investment will maintain a steady upward trend, with the five-year total investment scale increasing by more than 10% over the 14th Five-Year Plan period.

The second category is for advancing the green, low-carbon transition. This is mainly about implementing a ten-year doubling initiative for non-fossil energy. On the generation side, the dominant position of new energy will become more prominent, with its share of generation investment further rising to nearly 60%. On the grid side, we will build inter-provincial and inter-regional transmission corridors, inter-provincial power mutual-support projects, the grid’s main network framework, distribution grids, and smart microgrids, making the power system more resilient and improving the level of resource allocation. Investment in the grid during the 15th Five-Year Plan period will increase by more than 30% over the 14th Five-Year Plan period. After these projects are completed, the electricity everyone uses will be cleaner and greener.

The third category is for advancing the development of new-quality productive forces. This mainly includes laying out a batch of green hydrogen-ammonia-methanol production base projects, expanding multi-user direct green power connection scenarios, and building new-industry and new-business-model projects including new-type energy storage, integrated energy services, virtual power plants, and computing-power coordination. Investment in these new projects during the 15th Five-Year Plan period will exceed 2 trillion yuan—serving both as a “new engine” driving investment growth and opening “new space” for energy transition and development.

Energy project investment is large in scale and has a strong driving effect. This 20 trillion yuan invested will effectively drive the comprehensive development of the entire industrial chain—including scientific research and development, equipment manufacturing, and construction—injecting strong momentum into high-quality economic development.

Subscribe now

Just like other places around the world, China is facing the impact of AI. But if you talk with ordinary people, no matter whether they love it or not, almost no one disputes that AI is unstoppable. To understand the almost uniform posture Chinese people take toward artificial intelligence, you have to start from a point that many outsiders aren’t familiar with: 1840, the year the Opium War battered open the doors of a country that had sealed itself shut.

I know some American readers will see this and think, here we go, another round of the Chinese century-of-humiliation narrative. Yes and no. If you set the emotional storytelling aside, the real through-line of a Chinese high school history textbook is not about humiliation and revenge, but the Chinese people’s repeated attempts to save the nation, just the prescription constantly being swapped out. The Self-Strengthening Movement(洋务运动) tried to borrow the West’s machinery and technology, then failed at the First Sino-Japanese War. The Hundred Days’ Reform and the 1911 Revolution borrowed Western political institutions, ended up with warlords and chaos. The New Culture Movement borrowed Western thought and culture—and this thread doesn’t reach a resting point until the founding of the People’s Republic in 1949.

These attempts differed enormously from one another, yet they converge on a single conclusion that no matter which political system or which regime is in power, technological progress is the one thing you cannot get around. For anyone who has been through this Chinese high school education, a kind of progressivism about technology becomes deeply embedded in their view of history.

It’s precisely because of this that, for someone who grew up inside this basic education system and absorbed its historical narrative, artificial intelligence is simply another name for technological progress. And in this framework, technological progress belongs to the grand trend of the world—it is the direction history moves in, a current no one can opt out of. You either ride it or you drown in it.

Of course, education is itself part of class reproduction, so this logic won’t just stay in history books, it’s also conveyed to the country’s top-level institutional design. This June, Qiushi, the Party’s flagship journal, turned its attention to China’s reform of education, science, technology, and talent.

The popularization rate and key indicators of basic education in our country have comprehensively surpassed the average level of middle- and high-income countries. At the same time, however, problems still exist to varying degrees, such as placing excessive emphasis on the rote transmission of knowledge while neglecting the assessment of students’ comprehensive qualities—especially the evaluation of their innovative abilities—as well as ‘prioritizing selection over cultivation, and emphasizing the ‘creaming off’ of top talent over fostering innovation.’ We must promote the full-domain coverage of science and innovation education, stimulating students’ interest in innovation through everyday-life experiments, interdisciplinary projects, and the like, with a focus on cultivating students’ problem-solving abilities.

What this signals, on one side, is that China has recognized that the very basic education system that once underwrote its development is turning into its own opposite. In the AI era, the people that traditional education produces no longer meet demand, so education itself has to change to hunt for innovative talent. Seen from another angle, this amounts to once again defining the education system as a tool in the service of the goal of technological progress.

The Real Question Has Been Flipped Around

For exactly this reason, for many Chinese people, the real question was never why we should embrace AI, but on what grounds we would not. AI represents an advance in productive forces, and the lesson of history is that regimes that deny or suppress advanced productive forces mostly end up failing and collapsing. To put it more plainly: what people can do is adjust the relations of production to fit the development of the productive forces, but the productive forces themselves can’t be made to stop walking.

On this point, the Chinese logic resembles tech accelerationism; both accept that technological progress is unstoppable. The accelerationists of Silicon Valley are not a monolith—they range from a broadly left-leaning camp that dreams of a post-scarcity society of abundance, to the more radical “effective accelerationism” (e/acc) crowd, embodied by figures like Elon Musk, who want to harness ever-greater energy and turn humanity into a multi-planetary species. For neither group, however, is technological progress truly an end in itself; it remains a means to a more abundant human future. The crucial difference, then, lies less in the goal than in how and by whom technology is made to serve people. The accelerationist vanguard largely trusts that unfettered progress will deliver that better future on its own, whereas the Chinese view holds that technology should of course advance, but that the government simultaneously bears the responsibility for steering it toward the public good. The official vocabulary captures this with the phrase “AI for good” (智能向善).

Finance is one of the most concrete arenas in which the government is doing exactly this kind of guiding, and it was the theme of this year’s Lujiazui Forum, where the watchword was technology finance. The reason finance matters so much here is that, in this new wave of industrial revolution, the core industries represented by AI and semiconductors follow a developmental logic entirely different from the old land-and-credit model. On one hand, they demand enormous upfront capital; on the other, their asset-light nature leaves them with few physical assets to pledge as collateral. Combined with the heavy concentration of talent and capital they require, this makes tech firms far more dependent on, and far more demanding of, the financial system to feed them. Building a stronger and better-fitted financial system to support them has therefore become an increasingly pressing necessity.

How the very top is pushing this came through clearly in the speech that Wu Qing, chair of the China Securities Regulatory Commission, delivered at the forum. His core message was that capital markets must better serve new quality productive forces.(新质生产力) The concrete moves he laid out included extending the STAR Market’s fifth set of listing standards to the AI field, supporting the listing of high-quality AI large-model companies, and backing hard-tech firms in areas such as quantum technology, biomanufacturing and embodied intelligence, while also issuing, at an appropriate time, guidance to regulate the use of AI in the capital markets themselves. What gives the speech its weight is what it reveals that even the institutional plumbing of the capital markets is now making way for AI. Money, people, and institutions are all converging on technological progress.

Share

Are AI and Social Security Mutually Exclusive?

Of course, plenty of people argue that China going all in on AI means sacrificing people’s livelihoods in pursuit of technology. I think that claim is debatable. The way the Chinese government handles AI’s employment shock differs from the West’s, it leans more toward safeguarding job opportunities. captured the seeming dilemma and the growing policy push to secure jobs in his insightful piece, China is getting worried about AI & jobs. What I see in recent days is a growing number of policy advisers have been tapping the brakes on AI deployment.

One of the weightiest voices belongs to Jiang Xiaojuan, a former deputy secretary-general of the State Council. (Translation in my newsletter with her kind permission: Jiang Xiaojuan: AI’s Logic Must Not Prevail) Her judgment is that employment is a matter of people’s livelihood, and that at this stage we cannot yet let AI loose to allocate resources for the market. She singles out one trend in particular: in the past, technological progress created far more new industries and new jobs than it displaced, but since the 1980s that trend has been slowing. She therefore argues for cautious development of purely labor-saving AI, holding that AI deployment cannot for now be handed entirely to the market, and that government provision of a safety net is all the more necessary.

The stance of Cai Fang, a member of the People’s Bank of China’s Monetary Policy Committee, likewise reflects this Chinese view of technology. He doesn’t oppose AI; rather, he holds that technological progress can’t be reversed and that all people can do is adapt through reform. He sums up the principal contradiction in the current job market as a structural one—work with no one to do it, and people with no work to do—with AI playing the role of amplifier and accelerator, though not the only shock, and not the largest one. The prescription he writes out is “three big essays”: industrial guidance, individual adaptation, and social security. A few of his lines capture this adaptationism well. In an interview, he said that the AI dividend won’t automatically be distributed evenly to everyone, calling on the government to speed up and build a strong, multi-level social security system to create a solid safety net that protects people’s basic livelihoods. This will give working people enough time to learn new skills and change or reshape their careers.

Why China Doesn’t Want European-Style Welfarism

China’s social-security system is indeed on the weaker side; the pensions that urban and especially rural residents receive are quite limited. But to understand why China rejects European-style welfarism, I want to offer a discourse-level perspective.

Its origins trace back to the early stage of the reform and opening period. Before market-oriented reform can be advanced, a form of discursive preparation is usually required. The discourse of Reform and Opening condensed the ills of the planned economy into a set of highly negative keywords: egalitarianism(平均主义), the “big rice pot,”（大锅饭） low efficiency. This discourse helped China rapidly advance marketization, but at the level of popular culture it had a side effect that any institution tinged with universal provision and decoupled from performance got filed into this same category, so welfare became linked with laziness and waste. Set against this was the other side, jumping into the sea of business (下海经商), a striving culture of daring to dream, daring to act, daring to hustle, which was molded into a virtue. The negative imagination about welfare and the praise for striving are really two sides of one coin.

I also want to add another conceptual layer. In the Marxist understanding, labor and practice are the foundation of human society, and labor here is not equivalent to work in the narrow sense. You can love or hate your job, but labor is the process of creating and participating in social production, and it’s the way a person becomes connected to society. Once a person is severed from labor, severed from socialized mass production, they become disconnected from society itself. This is the deeper motive behind the government’s fixation on helping people find work: what it truly fears is not just that some people rely on welfare, but that a large mass of people will be cut off from any connection to society.

For exactly this reason, the State Council’s recently issued employment-first strategy in the 15th Five-Year Plan can be read as a head-on response to AI’s employment shock. It sets up an “action to adapt to AI development and promote employment,” which further breaks down into an AI job-creation plan, a plan to tap the employment potential of AI in traditional fields, and a plan to support workers in transitioning and switching jobs—the consistent logic being to reduce AI’s crowding-out of jobs and to amplify its job-creating pull. It also emphasizes using the service sector as a reservoir for absorbing employment, ensuring people have work while also improving the quality of that work by raising incomes, strengthening social insurance, and enhancing occupational safety.

A Short-Term Buffer, but a Welfare Still Has to Be Built

That said, the current toolkit is, at its core, still about balancing AI deployment against cushioning the short-term employment shock—what it’s buying is time. From a longer-term perspective, I’d still recommend the article Cai Fang published in Study Times, which Qiushi later reprinted. In it, he writes plainly that, taking universal basic income (UBI) as a reference point, China should raise the coverage and level of its minimum-livelihood guarantee, push the transition from a minimum-wage system to a living wage, and upgrade the urban-and-rural residents’ pension into a non-contributory basic pension that covers all elderly people unconditionally and equally. He also judges that AI-empowered robotics, displacing blue-collar work on a large scale, is the way the tide is turning. And once AI hits not just white-collar jobs but blue-collar ones too, the very premise that everyone has work to do can no longer hold.

My own read is that fiscal constraints will keep these goals off the near-term agenda. But the direction of travel seems clear. AI is quietly dismantling the old suspicion of the welfare system, and the welfare state, long treated in China as something to be kept at arm’s length, is turning into a problem there is no getting around.

Professor Huang has long been one of the most influential voices in China’s economic policy debates; he briefed Premier Li Qiang in 2023 during the economic situation symposium.

Huang argues that the very nature of China’s economic imbalance has changed. China has shifted from a small-country to a large-country economy; combined with rising trade barriers abroad, this means the old escape valve—exporting excess capacity—is closing fast. He suggests that expanding domestic demand has switched from a long-term goal to a short-term urgency. Signs of that are also shown in the difference between the 15th Five-Year Plan and the Central Economic Work Conference; the 15th Five-Year Plan puts forward two important tasks—developing new quality productive forces and expanding domestic demand—and the Central Economic Work Conference reversed their order, which suggested expanding domestic demand is now the priority.

He also calls for a fundamental rethink of how policy is designed: not a single stimulus lever, but a coordinated combination that simultaneously lifts demand in the short run and removes the structural roots of weak consumption over the long run.

Some other highlights:

• China is already rebalancing—just not enough. Huang pushes back against the common perception that no progress has been made. He points to consumption rising from 49.6% of GDP in 2010 to 57.1% in 2024, investment falling, and the current-account surplus dropping from its 2007 peak. Yet, with consumption still below 60% compared with a global average of around 75%, the work is far from done.

• The roots run deep. He traces “strong supply, weak demand” to a high savings rate, “asymmetric marketization” that suppressed factor prices for decades, local governments’ GDP-driven preference for investment over consumption, and a labor market that failed to translate growth into proportional gains in household income.

• A five-part policy combination. Huang argues that, unlike the Asian financial crisis era, strong macro stimulus alone won’t work, and lays out a coordinated package:

  • Macro policy should turn genuinely expansionary, but with the firepower aimed at households rather than at new investment projects—channeled through social security, welfare benefits, and direct transfers so that stimulus actually reaches consumers.

  • Market-oriented reform should finish the unfinished business of factor-price liberalization, correcting the “asymmetric marketization” that long held down the cost of capital, land, and labor at households’ expense, while restoring private-sector confidence.

  • Balance-sheet restructuring should address the strained balance sheets of local governments, property developers, and households, since over-indebted actors cut spending no matter how loose policy becomes.

  • “Investing in people” should expand public services and the social safety net—covering migrant workers through hukou and social-security reform in education, healthcare, and pensions—to lower precautionary saving and free up income for consumption.

  • A more responsible international role matters because structural contradictions should be resolved through domestic public policy rather than trade restrictions—yet other countries’ choices are beyond China’s control. With China’s export increase now drawing pushback, it must manage trade frictions rather than rely on overseas markets to absorb excess capacity.

• Income and confidence are the bottom line. Drawing on his own online consumer-brand research, Huang concludes that reviving consumption ultimately comes down to two things: people having money to spend, and daring to spend it.

  Share

Thanks Dr.Huang for authorizing me to translate his latest analysis into English. Below is the translation I made:

Policy Approaches Under the Macro Pattern of “Strong Supply, Weak Demand”

The Economic Growth Model

In his opening address at the Summer Davos Annual Meeting in 2007, Premier Wen Jiabao noted: “The overall situation of China’s current economic development is good. At the same time, there are also some problems in economic development that are unstable, uncoordinated, unbalanced, and unsustainable—mainly an excessively fast pace of economic growth, prominent structural contradictions, an extensive mode of development, excessive resource and environmental costs, mounting pressure of rising prices, and institutional and mechanism-related obstacles that have not yet been fundamentally eliminated.”

Twenty years have passed, and many of the issues now under heated discussion seem to bear similarities to those raised back then.

From the trends of the GDP growth rate and CPI in Figure 1, one can clearly observe phased changes. Roughly divided by time period: after the Asian financial crisis, the relationship between supply and demand began to exhibit the characteristics of “strong supply, weak demand.” Before that, China’s economy frequently faced high inflationary pressure; afterward, inflationary pressure weakened significantly, which to some extent reflects a shift in the supply-demand pattern.

Another key turning point appeared after the global financial crisis. Before it, China’s economy as a whole displayed the characteristic of being easy to overheat and hard to cool down; after the crisis, it shifted to being easy to cool down and hard to heat up, with the economy continuously facing downward pressure. At the policy level, growth-stabilization measures were introduced on multiple occasions, but after growth briefly stabilized for a period, new downward pressures would emerge again.

The core feature of the current macroeconomy is precisely this tendency to cool easily and heat with difficulty; viewed from the perspective of supply and demand, it is concentrated in the form of strong supply and weak demand.

Data on manufacturing capacity also corroborate the feature of strong supply and weak demand: China accounts for more than 30% of global manufacturing, ranking first in the world for 15 consecutive years. Of the world’s 500 categories of manufactured goods, China ranks first globally in production capacity for 220 of them. Looking at the manufacturing capacity utilization rate for 2025: China stands at 74.8%, the United States at 76.8%, Germany at 77.7%, Japan at 76.3%, and the global average at 78.2%. The gap between China and the major economies, as well as the global average, is not large, but overall it remains slightly on the low side.

The Causes of “Strong Supply, Weak Demand”

There are many reasons for the formation of the “strong supply, weak demand” pattern, and these reasons are not mutually exclusive or independent of one another—many are in fact interconnected and mutually influential.

1. High Savings Rate

First, the high savings rate is both an important cause of the current “strong supply, weak demand” pattern and, one could say, an intuitive result of this pattern. As Figure 2 shows, China’s gross national savings rate is at a very high level among the major economies. There is also one country with an even higher savings rate that is not listed—Singapore, at roughly around 50%.

If gross national savings is divided into the household sector and the public sector, one finds that China’s household-sector savings rate is indeed on the high side, at roughly 35% in the 2024–2025 data. However, compared with other East Asian economies, the gap is not particularly significant—for example, Indonesia’s household savings rate also reaches 35%, and many other economies are likewise at relatively high levels.

Some attribute this to Eastern culture’s reverence for saving and emphasis on long-term planning. I do not deny the influence of such cultural factors. But the high household savings rate clearly has other causes as well—for instance, the social security system is not yet well developed, so households need to rely more on their own savings to guard against future risks.

Although the household savings rate is high, the difference from other countries is not particularly pronounced. Another very important reason why China’s overall gross national savings rate is high is that the public sector accounts for a relatively large share of national income. For example, after a local government obtains a sum of land-transfer revenue, the proportion used for direct consumption is usually very low. Therefore, the public sector’s extremely high savings is an important factor pushing up China’s gross national savings rate.

2. Distortions in Factor Markets

More than a decade ago, we conducted related research and discovered a very distinctive phenomenon in China’s reform process, which we called “asymmetric marketization.” Since economic reform was launched in 1978, China’s overall direction has always been toward marketization, but the pace of marketization has been uneven across different domains. The core manifestation of so-called asymmetric marketization is this: the product market was largely and comprehensively liberalized fairly quickly, while distortions in factor markets remained relatively prominent. In the financial field that I follow, this feature is especially evident. The “financial repression index” shows that China’s index is significantly higher than that of most countries in the world, which to some extent reflects the considerable degree of government intervention that still exists in the financial system.

The fact that factor markets were not liberalized in step has multiple underlying causes, closely related to the gradualist, dual-track approach to reform that China adopted in the early period of reform. One important reason for adopting dual-track reform was to avoid the shock-therapy approach of liberalizing everything at once; the cultivation and refinement of market mechanisms is a gradual process, and abrupt, comprehensive liberalization tends to deviate from the intended objectives. The logic of the dual-track system is to retain the operation of the old system while liberalizing a new market track, gradually pushing the old track to converge with the new track and eventually merging the two. This model was once widely applied across many domains.

The advantages of this model are obvious: maintaining overall stability during economic transition, avoiding economic chaos caused by the sudden exit of the old system before the new mechanism has fully taken shape—this is crucial for a smooth transition. But it also has clear shortcomings: it brings about a certain loss of efficiency, and the coexistence of dual tracks easily breeds arbitrage behavior. How to continue deepening reform is a challenge we face over the long term.

Gradualist dual-track reform means that the old track needs to keep operating for a considerable period of time, and since the old track is inherently less efficient than the new track, ensuring its smooth operation objectively requires corresponding subsidies. And given limited fiscal resources, the most direct way is to suppress factor prices through distortions in factor markets, providing support in a disguised form. For example, in the early stage of reform, banks channeled low-cost credit on a large scale to state-owned enterprises—essentially a form of non-fiscal, implicit support.

In 2010, Tao Kunyu and I found in our research that factor-market distortions in 2009 were equivalent to 5.1% of GDP. According to a 2022 report by the team led by Gerard DiPippo at the U.S. think tank CSIS, which estimated the broad-caliber ratio of industrial subsidies to GDP (2019 data), China’s figure was 4.9% and the United States’ was 0.39%. Their estimates do not differ much from our calculations, but I do not agree with their simplistic characterization of attributing the entire amount to industrial subsidies. There is indeed a certain degree of industrial-subsidy behavior in localities’ investment-attraction efforts, but the underpricing of factors is, to a much greater extent, the cost paid for economic transition—an implicit burden-sharing during the transition process, rather than simply industrial support.

If one accepts this judgment of asymmetric marketization—that product prices were comprehensively liberalized while factor-market distortions have long existed and have not yet been fully eliminated to this day—then it means factor prices have generally been suppressed, which amounts to implicit support for producers, investors, and exporters over a very long period. And those who bear this portion of the cost are precisely the owners of the factors, primarily the household sector. This distinctive reform strategy has objectively formed a redistribution of income between producers and consumers. From this perspective, the “strong supply, weak demand” pattern has an intrinsic connection with the particularity of China’s reform strategy or path.

3. Local Governments’ Pursuit of Growth

As is well known, in the early period of reform China implemented decentralization reforms, devolving a great deal of resource-allocation authority from the central level to localities. This greatly mobilized localities’ enthusiasm for developing the economy and also allowed localities to formulate economic policies according to local realities. At the same time, competition among regions centered on GDP growth significantly boosted overall economic vitality, with local governments playing a key role.

For a very long time, local officials in charge were like CEOs of their regional economies, with their core work being to attract investment and advance construction. On the positive side, this did indeed powerfully drive economic growth—that is an objective fact; but at the same time, the tendency in the development process to emphasize investment and neglect consumption was also very prominent.

Consider this: after a local government obtains a sum of land-transfer revenue, is it more inclined to use it to drive consumption or to invest it in construction? Objectively speaking, in choosing the path of economic growth, stimulating consumption is slow to take effect and difficult, whereas investing in building infrastructure, creating industrial parks, and advancing industrialization yields quick results and provides tangible levers. This behavioral pattern was continuously reinforced, ultimately further intensifying the overall “strong supply, weak demand” pattern in the economy.

4. Wages Under a Labor Surplus

For a very long period in the past, China’s labor force was persistently in a state of surplus. In economies with abundant labor, the generally successful path to achieving economic takeoff is to start by exporting labor-intensive manufactured goods. In the first few decades of reform and opening up, we achieved tremendous success precisely by following this road. Not only mainland China, but also Japan, South Korea, and other Asian economies such as Taiwan, Hong Kong, and Singapore traveled similar development paths—that is, relying on abundant or even surplus low-cost labor to mass-produce labor-intensive products, forming significant competitiveness in international markets, and thereby driving export expansion and economic growth.

For a period of time, this model also brings about a typical consequence: because of the surplus supply of labor, wage levels are suppressed across the board. The economy was developing rapidly, yet wages failed to rise substantially in step, directly leading to the share of household income in national income falling rather than rising. This is also a phenomenon that many East Asian economies experienced in common during their high-growth stages—a declining share of household income and a correspondingly lower share of consumption in GDP. This pattern was also highly consistent with China’s economic structure at the time.

From the perspective of development economics, when the labor market shifts from surplus to shortage, the so-called “Lewis turning point” appears. Professor Cai Fang once judged that China roughly crossed this turning point around 2006. In other words, before 2006 the problem of labor surplus was very prominent, and after that this contradiction clearly eased. But now, when speaking of the job market, people’s intuitive feeling is that labor shortage is not prominent—indeed, quite a few people still face the pressure of unemployment.

5. Wage Income and Property Income

Over the past dozen or so years, innovations in digital technology—especially artificial intelligence technology—have to some extent had a substitution effect on labor. By the general rule, after the “Lewis turning point” labor wages should have risen substantially, thereby driving a significant increase in the share of household income in national income. But this phenomenon has not been evident over the past dozen years or so, and there are some special reasons behind it. Many trends that historically appeared in other countries have instead been less pronounced in China’s market—indeed, quite a few people still feel considerable employment pressure—which means that household consumption demand lacks strong income support.

There is also a view that, on the one hand, wage growth is weak, while on the other hand, against the backdrop of an aging population, society’s demand for property income is rising. In the future, as the labor force continues to shrink, wage levels will theoretically rise somewhat, but in addition to wage income, many people also need to rely on property income to support their livelihoods. If this portion of income cannot be effectively supplemented, it will likewise constrain the release of consumption demand.

6. Major Risk Factors

Major risk factors in economic operation have had a significant impact on the supply-demand relationship. Fluctuations in the real estate market affect the balance sheets of multiple economic sectors—including local governments, financial institutions, and the household sector—and their suppressive effect on aggregate demand, especially consumption demand, is very evident. The continued deepening of population aging puts long-term pressure on social security, the healthcare system, and government fiscal budgets. The strained finances of local governments have also, to a certain extent, affected infrastructure construction and the effective supply of public goods.

To summarize briefly: the current macroeconomy presents a “strong supply, weak demand” pattern, and the formation of this situation is the result of multiple factors working together. The several key causes just outlined can be summarized as follows:

First, the savings rate is high, with a correspondingly low share of consumption. Second, the past reform process had the characteristic of asymmetric marketization, with factor prices underpriced to a certain degree, further amplifying the “strong supply, weak demand” contradiction. Third, local governments have long been oriented toward GDP growth, and their tendency to emphasize investment and neglect consumption has aggravated this structural problem. Fourth, both before and after the “Lewis turning point” the labor market failed to drive significant and relatively rapid growth in household income, forming a clear constraint on consumption. Layered on top of these are multiple risk factors such as real estate, population aging, and local fiscal pressure. Overall, the feature of generally weak consumption demand in China over the past period has become very clear.

Already Rebalancing, but Not Enough

Many people have this question: Premier Wen pointed out problems with the economic growth model back in 2007, and in fact related discussions began even earlier—so why, even today, has the problem of structural imbalance still not been thoroughly resolved? In many people’s impression, the Chinese government has strong execution capacity and acts decisively, usually able to respond promptly when problems arise; yet on rebalancing and structural adjustment, no significant progress has been made for a long time. On this point, I do not fully agree.

Not long ago, the French presidency invited the Centre for Economic Policy Research (CEPR) to produce a study called the “Paris Report,” focused specifically on global economic rebalancing. France will host the G7 summit this year, and this report was prepared precisely for that purpose. In this report on global rebalancing, the organizers invited me to write the chapter on the rebalancing of the Chinese economy. The core argument I put forward first in the text is this: China’s economy has in fact already been rebalancing. Many foreign peers’ intuitive reaction to this judgment is one of disbelief.

But the economic data support my argument. The past economic structural imbalance, put simply, was that growth relied excessively on investment and exports while consumption was relatively insufficient. The indicator of this imbalance that draws the most external attention is the continued expansion of the current-account surplus and the trade surplus, with large amounts of domestic capacity being exported abroad.

Consider several sets of key data: First, the share of investment in GDP fell from 46.6% in 2011 to 40.6% in 2024. Second, the share of consumption in GDP rose from 49.6% in 2010 to 57.1% in 2024. Third, the share of the current-account surplus in GDP fell from 9.8% in 2007 to 2.3% in 2024. Although the share of the current-account surplus in GDP rebounded again after 2018, the ratio has remained at around 2% or below (3.7% in 2025).

At the time, the G20 Seoul Summit proposed a guiding range for external rebalancing—namely, that a surplus or deficit not exceeding 4% of GDP could be regarded as basically balanced. By this standard, China’s level in recent years has clearly been within the reasonable range. Since 2018, the share of China’s current-account surplus has rebounded somewhat, reaching 3.7% in 2025—a rebound from the previous few years, but still not touching the 4% warning line; this change, however, is worth watching.

Therefore, the rebalancing of China’s economy is in fact already happening, and great progress has been made compared with twenty years ago.

Of course, although rebalancing has made progress, it is still far from enough. At present, China’s share of consumption in GDP is about 57.1%, while the share of investment remains above 40%, and the contradiction of supply exceeding demand is still very prominent. From an international comparison, the global average consumption rate is about 75%, while China’s is less than 60%, indicating that consumption is still insufficient. Rebalancing still needs to be continuously advanced, and achieving a more balanced domestic supply-demand relationship in the future remains a major task.

Professor Helene Rey of London Business School once commented on my article, noting that China’s current-account surplus as a share of its own GDP is indeed declining—it has merely rebounded recently—but because the Chinese economy has grown so rapidly in scale, China’s surplus as a share of the GDP of the rest of the world or of its major trading partners may actually have risen. This means that, viewed from our own perspective, the relative proportion of external imbalance is declining; but from the trading partners’ perspective, the rise of China’s surplus relative to the size of other economies will increase the adjustment pressure on them. This is also a point we should take very seriously.

The Challenges of a Large-Country Economy

The outside world’s level of concern about our imbalance problem has not declined, and the core reason is that over the past twenty-odd years China has grown from a small-country economy into a large-country economy. No matter how a small-country economy’s imports and exports change, it can hardly affect the equilibrium of international markets; but having become a large-country economy, to put it colloquially, “whatever it sells becomes cheap; whatever it buys becomes expensive.” Even though China’s current-account surplus as a share of GDP has fallen and its own rebalancing has made progress, the rapid rise of its industries still brings tremendous adjustment pressure to other countries.

In early April 2024, then-U.S. Treasury Secretary Janet Yellen visited the National School of Development at Peking University, where she raised the question: with China’s new-energy vehicle scale so large, can the relevant U.S. industries still develop? She worried that large-scale capacity directed overseas would impact the U.S. auto industry, employment, and income distribution.

After World War II, when the United States led globalization, it advocated that countries divide labor based on comparative advantage, each doing what it does best for mutual benefit. But this set of ideas now faces challenges in reality, the key being that globalization is inevitably accompanied by structural adjustment. For example, China is good at manufacturing and Australia is good at growing grain, and trading with each other to exchange what one has for what one lacks is in itself win-win; but the problem that follows is: can Chinese farmers smoothly transition into manufacturing? Even if they can, the transition process is extremely difficult. The pain of such structural adjustment is a challenge that all countries face in common in globalization.

The U.S. problem is especially prominent. As an advocate of globalization and one of its biggest beneficiaries, the United States has a per-capita GDP that significantly leads other developed countries, yet it cannot escape structural difficulties either. Yellen specifically mentioned the employment plight of blue-collar youth in small American towns—something that has long gone beyond the economic sphere and evolved into a social and political problem. The emergence of anti-globalization policies in some countries is, to a large extent, precisely the externalization of domestic contradictions.

As an economist, my basic judgment is this: if a country concludes that globalization brings more benefits than harm overall, it should resolve structural contradictions through domestic public policy rather than simply resorting to trade restrictions. But the policy choices of various countries are not something we can control.

What needs to be made clear is that when an economy reaches a certain scale, changes in its supply and demand will significantly affect global markets and other countries’ decisions. Even if China’s current-account surplus share declines, it will still bring major adjustment challenges to many countries. In particular, China’s emerging industries are developing rapidly and its industrial chains are highly concentrated, so large-scale exports can very easily impact other countries’ industries. When countries generally worry about industrial security, employment stability, and income distribution, we, as a large-country economy, must consider how to develop in coordination with the world. A small country’s increased exports will be accepted, whereas a large country’s scale expansion will bring adjustment pressure—this is a new challenge we must face.

It is precisely for this reason that “strong supply, weak demand” has now become a prominent macroeconomic problem. But this is not a new problem. Since reform and opening up, overcapacity has appeared repeatedly. The former State Planning Commission (now the National Development and Reform Commission) issued warnings almost every year about overcapacity in certain fields and the need to avoid blind investment, and the governance of excess capacity also continued for many years—yet this did not prevent China from achieving high-speed growth averaging nearly 9% per year. The core reason is that in the past we were a small-country economy, and the domestic “strong supply, weak demand” gap was mainly absorbed through exports, forming an international grand cycle of “importing technology, capital, and raw materials from abroad, processing them, and exporting at low cost,” relying on external demand to absorb domestic excess capacity.

The situation now is this: on the one hand, the spillover effects of a large-country economy have triggered global vigilance, and other countries’ policy responses have clearly changed; on the other hand, countries are using trade measures such as tariffs to guard against market impacts. The room to balance domestic supply-demand contradictions by relying on exports has narrowed dramatically.

Although “strong supply, weak demand” is a long-term problem, its nature has fundamentally changed. In the past it was mainly a problem of structural rebalancing, a requirement for achieving long-term sustainable growth; now it directly bears on whether short-term growth can be stabilized. China’s capacity utilization rate is slightly below the global average, exports have remained strong in recent years, and the advantage of a complete manufacturing industrial chain is evident; but rapid export growth has simultaneously intensified external concerns—and not only in the United States and Europe.

The “15th Five-Year Plan” puts forward two important tasks—developing new quality productive forces and expanding domestic demand—and the Central Economic Work Conference reversed their order, with expanding domestic demand shifting from a long-term goal to a short-term and urgent task. If expanding domestic demand cannot take effect quickly and exports encounter obstacles again, large amounts of capacity may become substantively excess, in turn triggering risks to investment returns. General Secretary Xi Jinping published an important article, “Expanding Domestic Demand Is a Strategic Measure,” in the journal Qiushi, profoundly explaining the strategic significance of expanding domestic demand. How to effectively reverse the supply-demand imbalance is now the key to current economic work.

Shifting Toward the Domestic Grand Economic Cycle as the Mainstay—Developing New Quality Productive Forces and Expanding Domestic Demand

During this year’s Two Sessions, people noticed that the GDP growth target was adjusted from the previous “around 5%” to “4.5%–5%,” and I believe this adjustment is highly appropriate. Figure 3 is drawn from research by Professor Xiong Wei of Princeton University; the red line is the actual GDP growth rate each year, and the blue line is the annual growth target set at the Two Sessions.

In 2012 there was a clear inflection point: before that, the target was mostly set at around 8%, and the actual growth rate often substantially exceeded the target, making the task relatively easy to fulfill; after 2012, the growth target was continuously lowered, and the actual growth rate hugged the target tightly and closely.

In these years the growth target has had to be gradually lowered, and even so, ultimately achieving the target has often required going all out. And to ensure meeting the target for the current year, localities tend to adopt short-term measures that take effect quickly, and find it hard to lay out long-term reforms that are slow to take effect. This approach easily depletes future momentum and instead further intensifies downward economic pressure.

I strongly agree with moderately lowering the growth target, the essence of which is to leave room for transformation. I have recently been running with students and alumni, and I have a personal feel for this. Regular runners all know that starting too aggressively easily leads to running out of steam and crashing midway; only by slowing the pace at the beginning and laying a good foundation can you run longer and steadily pick up speed later. Economic development follows the same logic. From this perspective, this year’s target adjustment comes at exactly the right time. If we focus more energy on the two major strategic tasks—developing new quality productive forces and expanding domestic demand—and free ourselves from excessive pursuit of short-term speed, future economic growth will instead be more solid and more sustainable.

The new policy orientation is precisely to take the path of dual-circulation development, shifting from the past, where the international grand economic cycle was the mainstay, to the domestic grand economic cycle as the mainstay. Whether the domestic grand economic cycle can run smoothly hinges on getting two core links right: one is developing new quality productive forces, and the other is expanding domestic demand. Only by doing both of these solidly can we drive the economy to form a virtuous cycle, laying a solid foundation for the continued healthy growth of the future economy. Therefore, these two links are equally important; neither can be dispensed with.

Supply: The Evolution of Key Industries

Looking at the supply side, the process of China’s economic development over the past several decades has essentially been a process of continuous industrial upgrading and renewal. And the key question we now face is whether the next stage of industrial upgrading can be sustained and continuously advanced to higher levels.

Objectively speaking, China’s economic development has always been accompanied by innovation, but past innovation and current innovation are markedly different. Past innovation was more a matter of imitation and learning. For example, when we first developed manufacturing, many of the labor-intensive products we made were categories already matured by Japan and the Four Asian Tigers; it was only because their labor costs had risen substantially that the relevant industries gradually shifted to places like Shenzhen in China, and through following, learning, and rapid imitation, we achieved rapid industrial growth. One could say that innovation in that stage was at its core a process of learning and catching up. But now, China’s industrial development is increasingly approaching the global economic and technological frontier, and the demand for original innovation is also growing.

On whether China can continue to advance original innovation, there are differing views both internationally and domestically. I can clearly sense that the international perception of China’s innovative capacity has undergone a great change. In the past, many foreign experts dismissed China’s innovative capacity, believing China could only produce cheap products and could not deliver high-quality, innovative achievements. When speaking of Chinese innovation, the two characterizations they often cited were, first, theft of intellectual property, and second, reliance on government subsidies. But last year’s “DeepSeek moment” delivered an enormous visual and cognitive shock to many people.

An enterprise like DeepSeek neither stole others’ intellectual property nor received government subsidies, yet it developed highly competitive artificial intelligence technology and products. Looking back, there are in fact many high-quality innovative achievements domestically; not only has the international community underestimated our innovative capacity, but we ourselves often overlook it as well.

In fact, China already has quite a few emerging products that lead the world. More importantly, the current AI technological revolution offers us a rare historical opportunity, and the key going forward is to seize the opportunity and keep exerting effort, steadily advancing along the road of original innovation and driving continued breakthroughs in industrial upgrading.

Figure 4 is the global patent market share that I downloaded from a German research institute. In the figure, the red spheres represent China, and one can see that China’s patent market share in multiple fields is already in a leading position. Of course, the United States still maintains a commanding lead in many fields, but it is undeniable that China’s relative rise is very evident.

The Opportunity of the Technological Revolution

Some believe that we still have shortcomings in original capacity for technological innovation, with our advantages reflected more at the application level. Put simply, our breakthroughs from 0 to 1 are relatively weak—especially evident against the backdrop of China-U.S. technological competition—but in industrialization and large-scale application from 1 to 100, we do quite well. As an economist, I do not regard this as a shortcoming.

NVIDIA founder Jensen Huang once said: “The country that benefits most from a technological revolution is not necessarily the one that invented the technology, but the one that can apply it quickly and well.” We do indeed have an advantage in technology application, and if we can seize this round of opportunity and take application to the extreme, we can fully drive technological progress and economic development up a major step. On that basis, we can then gradually enhance our original-innovation capacity from 0 to 1 and compete with the world’s leading countries—this may be more pragmatic.

After all, we are still a developing country, and wanting to achieve original leadership in all technological fields overnight is unrealistic. Improving core technology and original capacity is itself a gradual process of continuous accumulation. And seizing the rare current opportunity to develop the economy well is the most crucial and most urgent task for us at present.

A Solow Paradox of AI?

In the process of the spread of new technologies such as computers, the “Solow paradox” once appeared: computers were everywhere, yet no obvious improvement in productivity was reflected. With the rapid development of artificial intelligence now, many people are also questioning whether a bubble has already appeared in the AI field.

Not long ago, when I spoke with Nobel laureate in economics Michael Spence, I also asked him this question. His view is that artificial intelligence does not count as a bubble, and even if a bubble exists, it is a rational bubble. Because when a promising new technology appears, no country and no enterprise wants to be left behind, and as long as real results ultimately materialize, the upfront investment cannot be counted as waste. The internet bubble at the start of this century is a classic example: when the bubble burst, large amounts of investment seemed wasted, but the boom of that era left behind a great deal of valuable technology and also gave rise to a batch of outstanding enterprises.

In the field of artificial intelligence, my colleague Shen Yan and others have done a preliminary analysis. From macro data, artificial intelligence has not yet had a statistically significant positive boost on China’s total factor productivity; but at the micro level, whether for enterprises or industries, the improvement of efficiency and productivity from AI can already be observed in real terms.

In other words, evidence of efficiency improvement at the micro level has already appeared, only it is still at an early stage; as for when effects will show at the macro level, that remains to be seen. If we draw on the pattern of how computer technology affected productivity back then, perhaps after some more time, artificial intelligence will play a tremendous role in improving overall efficiency.

Cathie Wood, an investor focused on emerging technologies, is very optimistic about the prospect of general-purpose technologies such as artificial intelligence driving economic growth, even believing that global economic growth could reach 7% by 2030. Whether this judgment is accurate is hard to determine, but what is certain is that after general-purpose technology is deployed at scale, it could fully drive economic acceleration and significantly improve efficiency—and this is a once-in-a-lifetime opportunity for us.

Therefore, from the supply side, the outlook is on the whole optimistic; there are already many successful cases of developing new quality productive forces, as well as many promising directions.

Demand: How Consumption Can Recover

“Strong supply” still has room for improvement, and “weak demand”—especially insufficient consumption demand—is the most important contradiction at present. Will China’s future economic growth be driven by consumption or by investment? There has long been debate on this. I believe there is no need to make an absolute either-or choice; maintaining an appropriate proportion between the two may be more important for a large-country economy.

Singapore’s savings rate is very high, yet this has not affected its continued growth. China’s current-account surplus as a share of GDP fell rapidly after the financial crisis from the historical peak of 9.8% in 2007, basically stabilizing in the range of 1%–2% after 2012, and rebounding to 3.7% in 2025. Singapore’s ratio, meanwhile, has been close to 20% year after year, yet no obvious problems have emerged—the reason being that it is a highly open small-country economy whose products are exported in large quantities and absorbed by external markets. But for a large country like ours, if the gap between the shares of consumption and investment remains too large, the difficulty in the future will only grow.

Boosting consumption has become crucial at this moment, for two main reasons. First, the ultimate purpose of economic development is to meet the people’s ever-growing need for a better life. A long-standing low share of consumption shows that many needs have not been truly met. Even though we emphasize long-termism and the accumulation of capital for sustained future growth, in the end we still must improve people’s lives, so raising consumption is entirely warranted.

Second, the more urgent pressure is that if consumption fails to rise, the “strong supply, weak demand” contradiction will become even more prominent. Against the backdrop of an increasingly complex international market environment and rising uncertainty in external demand, excess capacity will become increasingly prominent. Overcapacity means declining investment efficiency and wasted resources, and economic growth then becomes hard to sustain. In this sense, boosting consumption is no longer merely a question of medium-to-long-term structural rebalancing; it may well directly bear on whether economic growth this year and next can be stabilized—a very urgent, real-world problem.

Of course, it should also be noted that China’s consumption is in fact already recovering. The lowest point of the consumption rate appeared in 2010, and it has since risen by nearly 10 percentage points—only the pace of recovery has been somewhat slow and the process relatively difficult.

Over the past two years, the government has introduced quite a few policies, such as trade-in programs, and I have enjoyed the dividend of this policy myself. My old phone had been dropped beyond recognition, so I bought a new model, originally priced at nearly 10,000 yuan; under the trade-in policy, 3,000 yuan was deducted, and the experience was quite good. But the problem is that having bought a new phone this year, I will not buy a new one again over the next two or three years, even with a subsidy. The data also bear this out: in the early period after the trade-in policy is rolled out, demand is released in a short-term burst and rebounds quickly, but after one quarter it begins to weaken, and after three or four quarters it even turns into negative growth. Essentially, this pulls forward and over-draws future demand.

Therefore, fundamentally boosting consumption ultimately comes down to two points: first, having money to spend, and second, daring to spend money. The former is an income problem, and the latter is a confidence problem. All policies aimed at promoting consumption, if they are to truly form sustainable growth, must ultimately focus their efforts on these two points. Short-term stimulus policies such as trade-in programs are not without merit, but they can hardly form long-term, stable consumption momentum.

China Online Consumer Brand Index

Let me briefly introduce a piece of research that I have carried out with collaborators. Using data from across the Taobao-Tmall network, we constructed an online consumer brand index, the purpose being to provide an observable quality indicator for online consumer products. When we shop online, the easiest thing to see is price information, but there is no indicator for product quality. Take the used-car market, for example: for two cars both driven for eight years, ordinary consumers—mostly not experts—simply cannot see the differences in intrinsic quality, which leads consumers to ultimately tend to buy only the cheapest.

The theory of the market for lemons is a classic theory in economics used to describe market failure caused by information asymmetry; the lemon market is also called the market for inferior goods (”lemon” in American slang means “an inferior product, a bad car”). Put simply, in a market where the seller knows the true quality of the goods better than the buyer, the buyer can hardly tell good from bad and can only see the price; at that point, no one dares to buy the expensive ones and only dares to buy the cheap ones, with the result that quality products cannot be sold and gradually exit the market, and ultimately all that remains is inferior products, with the entire market getting worse and worse and even shrinking.

Some platforms repeatedly emphasize “the lowest price across the network,” even promising compensation if a lower price is found elsewhere. What impact will this orientation have on the structure of product quality in the market? All manufacturers are forced to make cheap goods, because the expensive, high-quality ones cannot be sold. Of course, lowering prices can compel enterprises to improve efficiency, but efficiency improvement has its limits; once the critical point is passed, pushing prices down further can only come at the expense of product quality. This is precisely the original intention behind our research: a market cannot have only price information; it also needs reliable quality information.

The Consumer Brand Index

Based on the data, we constructed two indices, one for goods and one for cities: one is the consumer brand index, which measures the average quality level of the consumer goods purchased by consumers in a city; the other is the consumer brand purchasing-power index, which measures the overall scale of spending by that city’s consumers on consumer goods.

As can be seen, the top ten cities in brand purchasing power are Shanghai, Beijing, Hangzhou, Shenzhen, Guangzhou, Chengdu, and so on, which basically accords with common sense—the more economically developed a city, the stronger its overall consumption purchasing power naturally is. Somewhat unexpectedly, the top ten cities in the consumer brand index are Hefei, Zhengzhou, Nanjing, Nanchang, Huai’an, Hangzhou, Linyi, Huaibei, Zhoukou, and others—these are not cities as developed as Beijing, Shanghai, Guangzhou, and Shenzhen.

Our analysis found two factors that can explain this. First, the higher the share of migrant workers, the higher the brand purchasing-power index and the lower the brand index. Large numbers of migrant builders have made important contributions to urban development, but their own consumption levels are relatively low, and they mostly buy low-priced, basic-model goods. The higher this group’s share, the more it drags down the city’s overall brand index. Second, the higher the share of labor within the establishment (i.e., the public/state sector), the higher the brand index. The income of this group is not necessarily the highest, but their security is more stable and their expectations clearer, so they have money and dare to spend it.

This shows that the key to boosting consumption still lies in the two fundamental issues of income and confidence. Making the economy larger in aggregate is only one aspect; the level of social security, residents’ income expectations, and the income-distribution structure are the foundations that support sustained consumption growth.

The Ranking of Up-and-Coming Fast-Moving Consumer Goods Brands

We have also studied quite a few up-and-coming fast-moving consumer goods (FMCG) brands, which I will not go into in detail here. What I mainly want to share is that from the brand rankings, one can clearly see that the products currently doing well are mainly of two types.

One type is traditional strong brands, such as Huawei and Apple, which still enjoy strong market acceptance even with relatively high pricing, because brand information is transparent and consumers have clear expectations of quality.

The other type that performs outstandingly is what we call up-and-coming FMCG brands, many of which come from emerging niches—such as baby sunscreen, pet products, and beauty cosmetics. Many of these subcategories were previously unfamiliar to the general public and even beyond many people’s imagination, yet they are all developing very well now.

What is even more interesting is that the manufacturing and supply-side innovation mentioned earlier is reflected especially prominently in the consumer goods sector. Of these successful up-and-coming FMCG brands today, more than 90% are domestic brands rather than international brands. And these domestic brands are priced no lower than international brands—some are even more expensive—yet they are still accepted by the market. This also shows that our innovation is diverse and rich, reflected not only in industries such as artificial intelligence, digital technology, and green energy, but equally in the consumer goods sector, where the deep integration of digital technology with market demand has given rise to a large number of entirely new products and niches.

The Future Direction of the Economy

The future direction of the economy is very clear, and the most important thing is “dual circulation.”

Among these, industrial upgrading and renewal is key. Over the past dozen or so years, our economic growth has continuously faced downward pressure, partly because industrial upgrading and renewal has not advanced smoothly enough. On the one hand, industries that have lost competitiveness and should have exited have not been cleared in a timely manner, and the overcapacity problems of many traditional industries remain prominent to this day; on the other hand, the scale and supporting strength of emerging industries are still clearly insufficient. Although emerging industries have developed rapidly in some fields, new problems have also arisen from everyone rushing in and expanding too fast.

Overall, industrial upgrading and renewal is both a huge challenge we currently face and a source of abundant opportunities. In my conversations with friends in the venture capital industry, they generally believe that another golden age of innovation and entrepreneurship has now arrived. When a general-purpose technological revolution arrives, the entire economic structure and business models may be redefined, and with that come a great many new opportunities.

These include opportunities brought by the rise of consumption demand in the process of domestic economic rebalancing, as well as opportunities to create new demand through supply-side innovation—many consumers originally did not know such products existed, but once good supply-side innovation appears, it quickly opens up the market.

From a macro perspective, we do indeed face enormous challenges, and economic transformation is by no means easy; but if we look at the micro level, we may well be standing at a new inflection point for a breakout, about to enter a period when opportunities for innovation and entrepreneurship emerge in concentrated form.

A Policy Combination

The current slowdown in economic growth is not simply a cyclical problem; it is not like responding to the Asian financial crisis in the past, where strong macro policy underpinned the economy and it could smoothly come out. The situation this time is not so simple and requires a policy combination, comprising five aspects: macro policy, market-oriented reform, balance-sheet restructuring, “investing in people,” and the international economic order.

1. Macroeconomic Policy

With the economy facing downward pressure, it is only natural for macro policy to exert force at the appropriate time, and I believe this year’s fiscal and monetary policy will have new arrangements. The difficulty is that macro policy must both stabilize growth and effectively ease the “strong supply, weak demand” contradiction—and if handled improperly, it may instead aggravate this contradiction. Because we have been better at stimulating supply in the past and not good at directly driving consumption, this is indeed a tough problem for decision-makers. Once stimulus is applied, funds easily flow into investment and production, making supply even stronger and demand relatively even weaker in the next stage; but the paths and levers for directly and effectively stimulating consumption are not clear.

Many experts believe that fiscal policy should play a more important role in this round of macro-control, because fiscal policy can more precisely reach sectors, industries, and specific fields, exerting force more directly. Of course, while increasing fiscal support, it is also necessary to balance the supply-demand structure—funds cannot be deployed only into infrastructure and industrial parks, or it will only make the “strong supply, weak demand” problem more prominent.

Monetary policy must likewise play a role, with the focus placed on effectively boosting demand. In these years the central bank has also been using and adjusting various structural monetary policy tools; the relevant measures must be precisely implemented to truly take effect, and the aggregate and proportions must be well managed—they cannot be overused, because monetary policy is essentially an aggregate policy. For example, now that “investing in people” is emphasized, could the central bank consider setting up a special re-lending facility for “investing in people”? If it could be genuinely implemented, the effect should be good.

Overall, macro policy must provide support for economic growth, but it may not be the most important role in this round of adjustment. When some countries roll out powerful macro policies, it is usually when the economy faces a serious crisis. Although we currently face downward pressure, we are still far from the stage where strong stimulus is needed.

2. Market-Oriented Reform

Even more crucial is to continuously advance market-oriented reform, allowing the market mechanism to truly play a decisive role in resource allocation—this is also the reform direction made clear at the Third Plenary Session of the 20th Central Committee. This reform is crucial to both the supply and demand ends.

From the supply side, having the market allocate resources means that investment and capacity decisions must follow market signals. The “involution” (neijuan) problem that has been hotly discussed in the past two years has complex causes, but we must be especially vigilant about local governments’ non-standard investment-attraction behavior. Such behavior easily deviates from market demand and expands capacity blindly; it fails to effectively raise technological levels and instead rapidly amplifies capacity, intensifying overcapacity pressure and depressing investment returns. Insisting that the market decide what to invest in and how much, while regulating industrial policy and restricting and clearing up non-compliant and illegal industrial subsidies, is precisely the clear governance direction at present.

From the demand side, deepening market-oriented reform helps market entities and the household sector obtain more income. For example, the land-transfer revenue obtained by localities is often concentrated by the government into large projects in pursuit of immediate GDP effects—equivalent to using all of it for investment, with a savings rate of 100%. If more public resources were directed toward people’s livelihoods rather than all concentrated on construction, residents’ consumption rate would naturally rise accordingly. Raising household income, improving public welfare, and having resources allocated more by the market, enterprises, and residents is more conducive to boosting consumption and to a virtuous economic cycle than allocation led solely by the government.

While advancing market-oriented reform, the public policy system must also be further innovated and improved. We currently have two core tasks—developing new quality productive forces and expanding domestic demand—and here lies a risk to be wary of: that new quality productive forces develop too fast while employment support fails to keep up, possibly further intensifying the “strong supply, weak demand” contradiction. For example, if artificial intelligence is deployed at scale and factory employment is greatly reduced, with residents’ wage income and job opportunities declining, where will consumption come from?

I once spoke with former U.S. Treasury Secretary Robert Rubin, who mentioned the employment plight of blue-collar youth in small American towns; China may face similar challenges in the future. As an optimistic economist, I believe technology destroys old occupations and also creates new ones. But Professor Jiang Xiaojuan’s observation has deeply inspired me: since the 1980s, the jobs and employment opportunities destroyed by technological innovation have exceeded the newly created positions, and the economy’s overall employment density has been declining.

Even as high-end industries flourish, the ordinary workers who are replaced find it very hard to transition smoothly. This poses a stern demand on public policy. Our technological innovation should adhere to an employment-priority orientation; technology should serve people, not simply replace them. There is nothing wrong with entrepreneurs pursuing efficiency, but the government needs to design new policy frameworks and tools to ensure that, throughout the process of technological progress and structural transformation, the people always have money to spend and dare to spend it.

3. Balance-Sheet Restructuring

The adjustment of the real estate market has already placed clear pressure on the balance sheets of households, local governments, and financial institutions, which has to a certain extent directly suppressed domestic demand, especially consumption demand. In this situation, consideration can be given to having the central government moderately add leverage to help various micro entities repair their balance sheets.

Although the central government also has some implicit debt, its overall debt ratio is relatively low, whereas local governments face heavy debt pressure, much of which is not directly reflected on the central government’s balance sheet. Local governments cannot go bankrupt, and once they encounter payment difficulties, the ultimate responsibility must still be borne at the central level. Therefore, through a round of systematic asset restructuring, the central government can moderately expand leverage to ease the balance-sheet pressure on other economic entities, creating conditions for economic recovery. This is exactly what the United States did during the global financial crisis. If large numbers of our economic entities continue to be troubled by debt burdens and shrinking assets, both domestic demand and investment will be severely suppressed, and the economy will find it hard to truly stabilize and rebound.

In addition, the balance sheet of the social security system is not very healthy either. At present, the social security fund as a whole still has a large funding gap, especially for the groups of rural migrant workers, migrant laborers, and rural residents, whose level of protection is on the low side and whose gap is more prominent. Although policy has been gradually improving in recent years, the magnitude of the annual increase is still on the small side. I believe consideration can be given to transferring state-owned assets more vigorously to replenish the social security fund and to carry out a more forceful structural adjustment of the social security system, genuinely strengthening residents’ level of protection and confidence to consume.

4. The “Investing in People” Policy

“Investing in people” is a direction that has been emphasized in recent years and is worth promoting with great effort. The fundamental purpose of economic development is to let the people live better.

“Investing in people” means improving the safeguards for people’s livelihoods across the full life cycle, including raising the quality of education, improving medical services, and refining the social security system, so as to enhance people’s sense of happiness and gain. At the same time, “investing in people” is also about accumulating human capital—by raising the quality of workers and supporting employment and entrepreneurship—to provide solid support for developing new quality productive forces.

In the future, whether local governments, various institutions, or society at large, all should shift their investment focus more from the traditional “investing in things” toward “investing in people.” This is conducive both to expanding domestic demand and improving livelihoods, and to effectively raising total factor productivity.

5. A Responsible Major Country

As a large country, the external environment and international rules we face are undergoing profound change, and we must participate in global economic governance in a responsible posture.

In recent years, discussions of international economic and trade rules have become increasingly intense, with some developed countries continuously questioning China’s status as a developing country within multilateral frameworks. In 2024, the U.S. Congress passed relevant legislation arguing that China should no longer be allowed to enjoy special and differential treatment as a developing country, contending that China’s economic scale no longer fits the relevant arrangements. In response, Premier Li Qiang announced at the UN General Assembly in September 2025 that China, as a responsible major developing country, will not seek new special and differential treatment in current and future World Trade Organization (WTO) negotiations. This important statement clearly demonstrates our sincerity and willingness to take the initiative in shouldering the responsibilities of a major country and to adapt to the transformation of global governance.

In the future, how to grow together with global economic partners for mutual benefit will be a major proposition we face. If we remain stuck in the traditional mindset of “my products are high-quality and low-priced and should rightly occupy the market,” our room for development will only narrow. Today many countries pay more attention to industrial security and structural adjustment, and pure advantages in price and quality are no longer sufficient to dispel external concerns.

In this regard, we are also actively exploring new paths. The Institute of South-South Cooperation and Development, established under the National School of Development at Peking University, is an important platform that General Secretary Xi Jinping advocated in 2015 and that was formally established in 2016. On the one hand, it distills pragmatic and feasible policy experience from China’s development practice, striving to form a “Global South consensus” suited to the broad range of developing countries; on the other hand, it seeks new approaches for enterprises going abroad, encouraging enterprises to deeply integrate into local economies—producing locally, creating jobs, and contributing tax revenue—thereby easing domestic capacity pressure while achieving an integration of interests with host countries.

A few years ago I proposed the idea of a “Global South Green Development Plan.” China’s green energy industry has significant advantages but also faces the pressures of overcapacity and trade barriers, while many developing countries urgently need green products to support their energy transitions. This is like America’s “Marshall Plan” of the past, which, through financial and industrial support, helped Europe revive—and the ultimate biggest beneficiary was in fact the United States, which both consolidated its alliance system and propelled the dollar onto the global stage. Such win-win cooperation is worth our careful planning.

From a domestic perspective, even if the “strong supply, weak demand” contradiction is hard to thoroughly resolve in the short term, as long as we continuously optimize the external environment and strengthen international cooperation, we still possess broad room for growth. In cooperating with many developing countries, competitive conflict is relatively mild—for example, our new-energy vehicles encounter resistance in some markets, but new-energy motorcycles are very popular in many African countries, where power infrastructure is weak, and we set up photovoltaic panels for them, solving the energy-supply problem. Overall, there is still much room for international cooperation. China’s economic development today is no longer simply about exporting products outward, but about growing together hand in hand with partner countries—this is the path of long-term sustainable development.

More to read:

The East is Read

Huang Yiping on China’s role in the Global South

Huang Yiping, Boya Distinguished Professor, Dean of the National School of Development (NSD), Dean of the Institute of South-South Cooperation and Development (ISSCAD) at Peking University, recently argued that China’s success model cannot be directly replicated by Global South countries and encouraged Global South students to adopt China’s pragmatism r…

Read more

10 months ago · 7 likes · Yuxuan JIA and Zhijian YAN

The East is Read

Huang Yiping on reduced local government roles and market-oriented reforms

Below is a speech at the National School of Development (NSD), Peking University, on September 9, 2024, by Huang Yiping, Professor and Dean of the NSD…

Read more

2 years ago · 6 likes · Yuxuan JIA and Andy Han

This brings to mind a piece in the book I’ve been reading. It was written in 2014, and it’s about how China’s wind power went from nothing to number one in the world. The funny thing is, the article is arguing about America’s “anti-dumping” case back then, yet the whole time I was reading it, what kept coming to mind were today’s European investigations and tools aimed at China. Twelve years on, the article isn’t just still relevant. It’s worth reading even more now.

This is the third episode of my series of readings of the book Pioneering Through Hardship: An Account of the Decision-Making and Construction of Landmark Engineering Projects (筚路蓝缕：世纪工程决策建设记述), the memoirs of Zhang Guobao, former Vice Chairman of the National Development and Reform Commission (NDRC). The title comes from the Zuo Zhuan 左传, an idiom describing the act of overcoming hardship to undertake a new venture under extremely difficult conditions.

Author Zhang Guobao served as Vice Chairman of the NDRC from 2003 to 2011, holding posts as the first Director of the National Energy Administration and as head of the leading group for the West-East Gas Pipeline project. He oversaw the formulation of industrial plans and policies for shipbuilding, automobiles, steel, and a host of other key sectors. More importantly, his tenure coincided with the pivotal eight-year window in which Chinese new energy shifted from a relative fringe option to a national strategy.

Reading this article on wind power, my biggest takeaway is that it’s really answering the question of whether China’s wind power competitiveness was built on subsidies or not. Zhang’s answer is a clear no. He believes the things that really did the work were three. Scale, localization, and competition.

The first is to make the market big enough. Only when the scale is there can you use the advantages of mass manufacturing to spread the cost of each turbine component over time. This is the fundamental reason costs later came down across Chinese wind power, and across the whole clean-energy industry.

The second is localization. Zhang mentions he once dealt with American middlemen who took a 200,000-yuan kickback just for selling a single turbine. With whole units relying on imports for years, costs simply couldn’t come down. The way he pushed localization was to go to the companies himself and mobilize them, piecing the whole supply chain together bit by bit, from blades, gearboxes and bearings to generators. The government’s role was more about coordinating and matchmaking than directly throwing money around.

The third is using competition to discover price. Through concession bidding, on one hand you can bring in competition among developers, and it also helps decision-makers grasp the real cost of wind power. As for private firms’ worry that state-owned enterprises would compete viciously with low prices and then monopolize the market, Zhang solved it by removing the highest and lowest bids and running multiple rounds of revision. And the facts proved it. The wind market is just too big, no one can afford rock-bottom prices time after time, so the so-called dumping and monopoly didn’t hold up.

Zhang also speaks directly about his view on subsidies. He agrees that in an industry’s early stage, it’s necessary for the state to give a certain amount of tax and R&D support, which I think is also a kind of “crossing the river by feeling for America’s stones.” But subsidies were never the reason Chinese wind power succeeded. While in office, Zhang strongly opposed the Ministry of Finance directly subsidizing domestically made turbines, believing direct government subsidies hurt market fairness. In 2009, when the U.S. accused China of industrial subsidies and sought to launch an anti-subsidy investigation, China took the opportunity during negotiations to cancel these subsidies. The clauses once used to protect domestic industry were likewise dropped voluntarily, after the NDRC judged that Chinese firms had the ability to compete head-on with their foreign counterparts. What really made Chinese wind power grow was still putting money into research, raising the industry’s competitiveness, and in the end bringing costs down through economies of scale.

Zhang doesn’t dodge the fact that, under China’s governance system, local governments, for the sake of tax revenue, would force wind companies to build plants locally, which eventually caused overcapacity. His attitude toward this, that “in the end it still has to come back to the market,” is actually a roundabout admission that the other half of this model is letting inefficient firms go bankrupt and letting the market clear. Scale and competition bring prosperity, but they also inevitably bring a reshuffle. The two are two sides of the same coin, and you can’t have it both ways.

Share

He also writes in detail about the delays in approving transmission lines and the arguments behind them, including whether to integrate the grid, whether to build outbound transmission corridors, and whether to go ahead with ultra-high-voltage. Everyone argued back and forth, but no one would make the call, and the result was that huge amounts of wind and hydro power couldn’t be sent out. Praise planning is easy, but it’s the willingness to talk about reality that’s convincing. The rest of the details I’ll leave to the translation below.

Honestly, while reading it, I often had a strange sense of déjà vu; the article talks about America’s “anti-dumping” case back then, but I kept thinking about Europe’s subsidy investigations into China. It’s reasonable for Europe to protect its own industries, but resorting to unilateral measures on Chinese goods does nothing, at root, to strengthen Europe’s competitiveness. By comparison, encouraging Chinese companies to invest in Europe would be a far more viable path. Yet Brussels’ approach brings to mind the satire of Yes, Prime Minister: just as British defence policy existed not to defend Britain but to make people believe it was defended, these measures exist not to protect European competitiveness, but to make people believe it is being protected. Worse still, they risk inviting Chinese retaliation that would only deepen Europe’s economic woes. And this article is itself the answer: China’s wind power succeeded not by relying on subsidies, but on scale, competition, and research. Protection might buy a moment’s breathing room, but it can’t buy real industrial competitiveness. And this holds true whether for the China of back then or the Europe of today.

Below is the full piece:

The Decision-Making Behind China’s New Energy

In ten years, China’s wind power industry leapt from an obscure sector to a position of relative global leadership. “First, make the ‘pie’ bigger; second, introduce competition; third, push for domestic manufacturing—these enabled the industry to achieve explosive growth.” But amid this explosive growth, numerous controversies also arose, such as “wind curtailment,” overcapacity, excessively high costs, the form of subsidies, and whether to pursue concentrated development or distributed development. After 2009, worries, criticisms, doubts, and even accusations about wind power’s “Great Leap Forward”–style development came in an unending stream. To sum up our experience and draw lessons, we should not shy away from the problems, and it is necessary to review the decision-making and development course of new energy over these ten years.

Three Major Policies Sparked Explosive Growth in Wind Power

The application of wind power in China can be traced back to the 1980s, but it was not until the “Eleventh Five-Year Plan” period that it ushered in explosive growth. Wind power did not appear only in recent years. Before I served as Deputy Director of the State Planning Commission, my predecessor as Deputy Director, Ye Qing—who later became Chairman of the Shenhua Group—was someone I once accompanied to the United States for an inspection visit, probably in the early 1990s. When the plane landed in California, we saw wind turbines covering the hilltops along the coast in great swaths.

New energy in China got started as early as the 1980s. In the 1990s, the predecessor of today’s National Energy Administration was the Basic Industries Department of the State Planning Commission, and within the Basic Industries Department there was a division called the Energy Conservation and Renewable Energy Division. The function of this division was to develop new energy using international loans. At that time the World Bank provided a sum of money and set up a dedicated office to promote wind and solar energy, called the World Bank Renewable Energy Loan Office. The Energy Conservation and Renewable Energy Division had also at that time selected manufacturers such as the Luoyang Tractor Factory and the Xi’an Aircraft Manufacturing Factory to produce wind power generation equipment, but none of them later amounted to much. At that time, wind power really had not developed on a large scale. In 1999, I took over as Deputy Director of the State Planning Commission, also overseeing energy. The entire country’s installed wind power capacity by 2000 was still less than 400,000 kilowatts, and this 400,000 kilowatts of equipment was basically purchased from abroad, mainly Vestas and GE equipment.

After 2003, renewable energy represented by wind power did indeed usher in explosive development, for a variety of reasons. Looking at the broader situation, worldwide the green-development concept of new energy was increasingly winning people’s hearts, and the same was true in China. The state’s emphasis on renewable energy was gradually strengthening, and public awareness was also continuously rising, which provided a very good public-opinion environment for promoting renewable energy. On the other hand, there were the state’s guidelines and policies. After I took office, I felt very puzzled: why couldn’t wind power be promoted? Later I learned that wind power was extremely expensive—the cheapest electricity price was still eight mao per kilowatt-hour, and some even reached around two yuan per kilowatt-hour; naturally, not many people were willing to use electricity that expensive.

At first I did not understand: wind resources don’t cost money to buy, so how could the cost be even higher than that of coal-fired power plants? It turned out that the main cost was in the equipment—the price per kilowatt of wind power equipment was much higher than that for thermal power, and part of the equipment investment was financed by loans. Although the wind itself costs nothing, the principal and interest repayment, equipment depreciation, and other financial expenses were the main costs, so the price of wind power was extremely expensive. Therefore, I put forward suggestions in three areas:

First, make the “pie” bigger. Since it was so expensive, we had to find ways to make it cheaper. How could it be made cheaper? By making the “pie” bigger and spreading out the cost. If there were only three to five machines and no large-scale production, the cost could not be brought down. If each manufacturer made several hundred or a thousand units, the cost would be spread thin, and enterprises would have an incentive too.

Second, push for domestic manufacturing. Why was it so expensive at first? Because we couldn’t make it ourselves. When we first did research, we encountered many middlemen. Earlier, when I went to the United States with Comrade Ye Qing, we ran into two ethnic-Chinese brothers doing this kind of business. They sold American wind turbines into China. It was said that the kickback for selling a single turbine was 200,000 yuan. This was how equipment prices were driven up. If we could solve the equipment problem ourselves through domestic manufacturing, prices would come down.

Third, introduce competition. In the past there was no competition among wind power developers. If a certain place had wind and the government wanted to build a wind farm, and a developer was willing to invest, the project would be given to him. Once it was up and running, he would go lobby the government, getting the price bureau to approve the electricity price, and some were approved at 1.5 yuan/kWh or even 2 yuan/kWh. Investors naturally hoped the price department would approve a high electricity price, because the higher the price, the more they could earn. At that time, investors and project owners were determined in this way. Later, I felt this should be changed: the government should not designate who would be the investor; those willing to invest in wind power could compete, conducting concession-rights bidding. The government would provide wind resource data to investors, and the investors would bid against one another on price, determining how much per kilowatt-hour. This created competition: there might be three to five parties all wanting to do it; one would bid six mao per kWh, another five mao five per kWh, and yet another five mao per kWh—then I could let the one who bid five mao per kWh do it.

Through these three methods, we brought down the construction cost, and the price per kilowatt-hour came down as well.

The Concession-Rights Bidding Debate

The practice of concession-rights bidding stirred up considerable debate at the time. Originally, electricity prices were all determined by the price department based on cost plus an appropriate profit. They still preferred to do it this way and did not want bidding, saying that after bidding there would be vicious competition, leaving wind power enterprises with no profit to make, which would dampen the enthusiasm of wind power investors and possibly have a bad effect. Others said that state-owned enterprises could disregard cost and maliciously drive down prices in competition, squeezing private enterprises out. This was what they considered the problem with concession-rights bidding. In addition, they raised the issue of malicious bidding: because if you bid five mao, then in order to beat you I would bid four mao—when actually four mao means a loss and can’t be done—this would dampen everyone’s enthusiasm for investing in wind power and distort the normal bidding price, and so on. At the time there were some such reasons for opposition.

Would the situations they described actually happen? Yes—sometimes bidding might indeed give rise to vicious competition. But we could do as a referee scores points: to avoid unfairness, drop the highest score and the lowest score, take a weighted average, and on the whole it could still reflect the true level. Moreover, even if someone competed viciously and deliberately took a loss, that kind of thing cannot last—once is fine, but can one keep taking losses two or three times? Some also charged that private enterprises would surely withdraw because they couldn’t afford the losses, while state-owned enterprises could compete without regard to cost. Actually this is wrong too. Don’t assume state-owned enterprises will keep operating at a loss forever. An enterprise might engage in vicious competition once, but it cannot do so every single time.

Why did the concept of concession-rights bidding come about? In the past I served as head of the leading group for the West-East Gas Pipeline construction project, and at that time we also used bidding to determine the supplier of pipeline steel. Back then we also pushed for domestic manufacturing; the domestic firms that could make it included several large steel companies such as Baosteel, Angang, and Wugang. At this point a South Korean company came in and bid lower than all the domestic firms. Many domestic steel mills came to complain, saying this was vicious competition—such low prices were dumping—but at that time there was no concept of anti-dumping yet. As the owner unit, the responsible comrade from PetroChina came to me. He said, “Director Zhang, what do you think we should do? Whom should we let win the bid?” At that time I was also in a difficult spot: the domestic steel enterprises reflected that the Koreans were competing viciously, but the owner unit’s interest was oriented toward buying from whoever was cheapest. All right then—the first time, let the Koreans win the bid. The Koreans were indeed competing viciously, dumping at below cost, but at the second bidding, that Korean bidder had no way to continue, because he had done a loss-making deal. At the second bidding, it was all Chinese enterprises that won.

Judging from this experience, I believe the logic of wind power concession-rights bidding is the same: even if someone competes viciously and bids recklessly, that situation cannot last. A similar incident also occurred during the first batch of wind power concession-rights bidding. In 2003, we held the first batch of wind power concession-rights bidding in Rudong, Jiangsu. At that time I was away on a business trip; Shi Lishan, now Deputy Director of the Renewable Energy Department, was then a division chief, and he called me and said: “Director Zhang, there’s a company called the Huarui Group, a private enterprise, and it bid only three mao eight per kilowatt-hour. Everyone who participated in the bidding thinks this is impossible to achieve. But it bid three mao eight, while everyone else bid five mao—what do we do? Whom exactly should we let win the bid?”

According to the bidding rules, since he bid the lowest he should be allowed to win, but everyone thought three mao eight was impossible. This stumped me. I could only say, just let him win the bid. I knew it was indeed false and he surely could not carry it through. It was later confirmed that he indeed could not carry it through, and construction progress remained extremely slow throughout. After this kind of thing happens once, it is very hard for it to happen a second time. Throughout this whole affair we had to keep correcting course and summing up experience; from time to time various situations would arise, but a normal enterprise cannot do loss-making business forever, and competition can still push everyone to lower costs and improve technology. I have always firmly believed that competition should be introduced, as this is somewhat more scientific than having government officials set prices arbitrarily.

To sum it up in my own words, this is called discovering the price through competitive bidding. In a given region, through several rounds of bidding and several corrections—just like dropping the highest and lowest scores—one can gradually discover a true price, at what level it is reasonable. As it now appears, it is roughly between five mao and six mao, and this price level is fairly appropriate—neither yielding windfall profits nor incurring losses.

However, the price department, which advocated government price-setting, still did not change its view and kept criticizing concession-rights bidding as no good, so later I compromised too. Now that bidding has discovered the reasonable price level, we have formed benchmark feed-in tariffs for four categories of regions—setting one benchmark tariff for regions with roughly similar wind resource conditions—so we no longer hold bidding, but set prices according to the approximate data from the previous rounds of bidding. The country is now divided into four categories of regions: 0.51 yuan, 0.52 yuan, 0.57 yuan, and 0.61 yuan. If you want to build a wind farm in a given place, you go by that region’s electricity price.

Although we have now returned to determining electricity prices according to each region’s resources, this process discovered the price through competition, and in this process it rapidly lowered the cost of wind power—from a high of 0.8 yuan or even 2 yuan, down to roughly between 0.5 yuan and 0.6 yuan now.

Some say that in this process private enterprises couldn’t make money and all withdrew. Practice has proven this is not the case. Not long ago a private enterprise even came to me, hoping I would do some work to get the Energy Administration to approve their wind power project. I also specifically asked the major wind power investors about this matter; I said, are you actually losing money or making money? Longyuan Company made 3 billion yuan last year. During the 2014 Spring Festival I also ran into He Yu, Chairman of CGN, and he said CGN was also making money in wind power, and the same was true of Hebei Construction & Investment. So it is not the way some people imagine—that state-owned enterprises compete viciously without regard to cost; if no one were making money, they could not survive forever.

Domestic Manufacturing Drove the Growth of a Group of Enterprises

In those years, several academicians working in aerodynamics, in the process of promoting domestic manufacturing, used their technology and expertise to establish AVIC Huiteng Wind Power Equipment Co., Ltd. in Baoding, going into business themselves to make wind turbine blades. There was also a new-materials company, spun off from the Building Materials Bureau, which used its strength in making fiberglass to produce fiberglass blades in Lianyungang, Jiangsu. No one had been able to make wind turbine gearboxes either, because the technology was very complex; only the Chongqing Gear Factory produced some, drawing on the capability it had formed making marine gearboxes. I thought of the Dalian Heavy Machinery Factory and the Nanjing High-Speed Gearbox Factory, and I suppose it was I who egged them on. I said wind power was about to enter a very good period, and if they got involved, they would find this was a very good market. I did some promoting to them, but in the end they themselves made up their minds to do it. At first they were rather conservative, saying they’d make a thousand units. I said a thousand is too few—make at least two or three thousand. The Nanjing High-Speed Gearbox Factory had been restructured into a joint-stock company; in the late period of the “Cultural Revolution” it had worked on the project to transmit Sichuan gas out of Sichuan. Sichuan’s abundant natural gas could be exported for gas-fired power generation, so a gearbox factory was set up. This gearbox factory made gearboxes for gas turbines, but later there weren’t many orders, so it switched to making cement and marine gearboxes, and still later it made wind power gearboxes. When I went to inspect that factory, I said to its leaders that if they got into wind power now, they’d have at least five to ten years’ worth of work to live on. Later they did it.

Once gearboxes were available, bearings still couldn’t be made and were all imported. At that time I was presiding over the revitalization of the Northeast, and there was the Wafangdian Bearing Factory locally. This was an old state-owned enterprise that originally made ordinary bearings, but now it needed new products to compete with private enterprises. I said what private enterprises can’t make at present is the large slewing bearings for wind turbines, so they too began to make them. There was also the Jiamusi Electric Machinery Factory, which had previously made explosion-proof motors, and I introduced them to making wind power generators.

Although to this day some components still cannot be domestically manufactured, the process of an industry going from being completely unable to manufacture domestically to manufacturing a considerable portion domestically not only built up the entire supporting supply chain but also drove the growth of a large group of enterprises.

Subsidies Are Not the Fundamental Solution

In the early stages of renewable energy development, quite a few countries provide subsidies to the industry, and quite a few people called on our government to provide subsidies. But I believe subsidies are not the fundamental solution to the problem. So among the policies I mentioned, I did not particularly emphasize providing subsidies.

One substantive subsidy was when we obtained some money from the Ministry of Finance at the time, about 80 million yuan, all of which was given to the Meteorological Administration to measure wind and provide nationwide meteorological data. Another item was support for research and development including wind power. At that time I asked the Ministry of Finance for some money, 200 million yuan in total, but not all for wind power—rather, it was used to build key energy laboratories, including the national wind power laboratory and the offshore wind power laboratory. This 200 million yuan was used for laboratory research facilities, and that counts as a subsidy too.

There is one more point, which is the electricity-price subsidy still in effect now. Now wind power is at minimum five mao and at most six mao one, roughly within this range. If the thermal power on-grid price is four mao five and wind power is five mao five, a difference of one mao, that price-difference portion is subsidized by the state through the renewable energy fund. Where does the renewable energy fund come from? A levy of eight li per kilowatt-hour. People nationwide pay eight li per kilowatt-hour. Some don’t pay—for instance, captive power plants and rural small hydropower don’t pay. All the remaining large power plants pay eight li per kilowatt-hour, which can collect 20 billion yuan a year to subsidize renewable energy. But I still have some worries about this subsidy method: if the scale gets larger in the future, more subsidy will be needed, and eight li won’t be enough.

It can be said that during the growth of the wind power industry, the state’s substantive financial subsidies were actually not much; it was mainly through the industry’s development prospects and policy guidance that the industry was driven forward.

New energy is a new concept, a good concept; not only were we at the Energy Administration and the NDRC willing to give support to wind power, but the Ministry of Finance and the Ministry of Science and Technology were also willing to make some contributions. The Ministry of Finance once issued a policy: for the first 50 domestically manufactured wind turbines of 1.5 MW and above, a subsidy of 600 yuan per kilowatt. The Ministry of Finance once sought my opinion on this matter. At the time I disagreed. I said this kind of subsidy is somewhat unfair—some get the subsidy, some don’t—and it would be better to use it to support wind power R&D. But the Ministry of Finance carried out the policy anyway, giving 600 yuan per kilowatt for the first 50 turbines, for a cumulative subsidy of 200 million yuan. Later this matter was seized upon by the Americans, who accused the Chinese government of providing subsidies and conducted an investigation against us similar to “double anti-dumping/anti-subsidy.” When I learned of this, I proposed holding a video dialogue with the Americans for everyone to discuss. Because the United States subsidizes new energy far more than we do—you can look up on websites the new energy subsidy policies issued by each state, which are simply as numerous as the hairs on an ox—they had no grounds to turn around and accuse us. The Americans agreed, but they did not allow the media to participate; it could only be official-to-official dialogue, and the participating departments were the U.S. Department of Energy, the Department of Commerce, and others. The U.S. Department of Energy was actually worried about coming into conflict with us, and repeatedly explained to us that this was done by the Department of Commerce and that the Department of Energy was just a foil. On our side, the lead was me; at the time I was Director of the National Energy Administration and the Deputy Director of the NDRC in charge of energy.

I also deliberately put them on the spot a bit by saying I wanted to bring in the media. I said, don’t you Americans advocate transparency? But when it really comes to being transparent, you get scared—to put it in a Chinese idiom, it’s like Lord Ye loving dragons [professing to love something one actually fears]. Their reasoning was that having the media present might make the problem harder to resolve and would only keep hyping the matter. At the start of the dialogue the American side asked, I’d now like to ask our Chinese friends to confirm—on your side there’s no media, right? I said I could confirm there was no media. In the dialogue I listed out the various American subsidies, and after going back and forth, the Americans had no truly tenable evidence, and in the end they latched onto the one item—the Ministry of Finance’s subsidy at the time of 600 yuan per kilowatt for the first 50 domestically manufactured wind turbines of 1.5 MW and above. Later the China-U.S. Joint Commission on Commerce and Trade was held in Hangzhou, led by Comrade Wang Qishan; the U.S. side was led by the current Ambassador to China, Gary Locke, who at the time was Secretary of Commerce. We assessed in advance that he would surely raise this issue of wind power subsidies.

Later I suggested to Comrade Wang Qishan that if he raised this matter, we should take the initiative to say we’d cancel this subsidy. After communicating with the various departments participating in the Joint Commission, the Ministry of Finance also agreed with our view. At the Hangzhou meeting, sure enough, Gary Locke raised this issue. Comrade Wang Qishan then said, all right, we’ll cancel it on our own initiative. Locke then had nothing more to say and could only say we are very pleased that this meeting was so productive. Later the U.S. “double anti-dumping/anti-subsidy” investigation against China’s wind power never got off the ground.

In the early development stage of an industry, the state should provide support, but such support is best placed in R&D or taxation. In fact the United States is the same: they place their support policy for wind power in taxation—not taxing or taxing less, using tax incentives as a form of subsidy rather than directly giving funds. If you directly give financial subsidies, then the larger the future development scale, the larger the subsidy scale, and at a certain point the public finances cannot bear it. So I advocate that the wind power industry should still take the development path of enhancing competitiveness, raising R&D levels, and lowering costs. There is one more point, which is to make the “pie” bigger; relying entirely on state subsidies cannot last—this has always been my philosophy.

On this development philosophy of mine, the various stakeholders did not hold uniform views. The price-control department did not agree with the bidding approach; they believed it was still best for the government to determine a price through cost plus appropriate profit. I disagreed. I was Deputy Director of the NDRC; I did not directly oversee prices, but I was also their leader, and it was awkward for them to openly defy me, so they still agreed—but in fact they always held reservations. After I was no longer in charge, they reverted to setting by region and no longer competed.

This view of mine encountered fairly large resistance in the course of implementation, and there were some critical junctures that made me feel rather strained in carrying it forward. At the time some interest groups voiced their feelings through the media, saying this approach of mine was unscientific. As mentioned before, some held that bidding might cause vicious competition, that state-owned enterprises disregarded cost, that private enterprises had no way to get in and no money to earn, and that in the future no one would be willing to invest in wind power—there was a great deal of such public opinion. But the facts proved otherwise: in these very years wind power achieved great development. This confirms that practice is the sole criterion for testing truth, rather than relying solely on what certain people imagine in their heads.

We originally had another policy: in order to encourage domestic manufacturing, there was a scoring condition in the bidding—a localization rate of 75%; if you reached 75% you could earn extra points, and if you didn’t reach 75% it wouldn’t do. The Americans also latched onto this against us. Later, since we already had three to five years of practical experience, we judged that even if this condition were removed, Chinese enterprises should have the ability to compete on the same stage as their foreign counterparts, and the state cannot adopt protective policies forever. So after the China-U.S. Joint Commission on Commerce and Trade in Hangzhou, we canceled this condition, and the handle the Americans held was gone.

There had also previously been a stipulation on operating qualifications: at the time of bidding, enterprises were required to have an operating track record of more than 100 turbines within China; later we also recognized bidders’ operating track records abroad. The United States later also used this clause to constrain us, saying that for China’s wind power equipment to be exported to the United States, Chinese companies would be required to have an operating track record of more than 100 turbines on U.S. soil. I then said, back then you used this very issue to accuse us, and now that we’ve canceled it, you instead start using this clause. We demanded that the United States should recognize manufacturers’ track records achieved elsewhere in the world, not only allow it if they had such a track record on U.S. soil. So there are all kinds of games being played in here.

In the End, We Must Return to the Market

Looking at the course of new energy development to date, the rise of both wind power and photovoltaics has been closely tied to local development. The relationship between the new energy industry and local governments over these years is also an issue in renewable energy development. Originally, local governments were quite enthusiastic about developing new energy, because many wind power locations were often barren land where not a blade of grass grows, with no income per unit of land area, but developing wind power could bring in some tax revenue and yield some returns, so local governments were supportive. Later the Ministry of Finance introduced a policy called equipment deduction; the starting point was good—to encourage everyone to invest in wind power, the money spent buying equipment to invest in wind power could be deducted from future taxes payable. This measure sounds at first like support for wind power development, and as a wind power investor one would certainly be glad, since it lightened the tax burden. But local governments’ tax-revenue interests were affected—they had no tax revenue for many years. Having lost the tax revenue, local governments began to ponder a question: why should I invite you to come and do this? What benefit can I get? Now I get no benefit, and the tax revenue is gone too. What is a local government to do? It could only force enterprises to build factories locally. For example, a certain city might have wind resources and be able to build a wind farm, but the government would require that wind turbines produced in that city must be used, thus forcing these turbine manufacturers all to build factories locally. The underlying cause behind many problems between renewable energy projects and local governments is taxation.

Regarding the relationship between renewable energy enterprises and local governments, one can look at today’s photovoltaic industry. Companies like LDK and Suntech were once leaders in the solar industry; in fact, neither the NDRC nor the National Energy Administration gave these enterprises many real benefits. When I was Director of the National Energy Administration, the national finances gave them no subsidy. But local governments believed these companies provided a relatively large boost to the local economy, so it was inevitable that they would offer all sorts of incentive policies. Most of these former leaders grew up amid the great tide of the market economy; they enjoyed the preferential conditions provided by local governments, and now, due to overcapacity and the squeeze on us from foreign-market protectionism, they are going through a very difficult period.

From this I also came to realize a principle: the development of an industry and of enterprises must rely more on market forces. In the startup phase, it is also wrong for the government to give no policy at all and no encouragement, but in the end one must return to the market and enhance one’s own competitiveness—only then can one maintain vigorous vitality over the long term.

Under the current development situation, the existing subsidy mechanism, or more broadly the entire support policy, has also reached a point where it needs further improvement. Enterprises often, on the one hand, oppose the planned economy, and on the other hand, hope the government will give more subsidies. In the long run, giving them subsidies is like giving them medicine—once the medicine stops, they have no resistance. I do not advocate giving subsidies long-term; rather, I advocate that the government make the “pie” bigger. Making the “pie” bigger may be more effective than giving subsidies, and the real financial subsidy should be applied to R&D.

The same goes for the renewable energy fund: once the scale gets large in the future, it surely won’t be able to subsidize so much, and by then even levying one fen per kilowatt-hour might not be enough. The most fundamental way out is still to bring down costs through market competition and technological progress, down to a level comparable with fossil energy—this future should be foreseeable. The problem now is that adopting an arbitrary price-setting method has shackled the motivation to keep lowering costs. Once you’ve set it at six mao one per kilowatt-hour, it’s indeed convenient and enterprises have profit to make, but then there is no incentive to pursue lower costs. I think we should still have competitive bidding, so that enterprises have the incentive to further research new technologies.

The Key Crux: Lagging Approval of Transmission Lines

In the years since the Renewable Energy Law was implemented, it has frequently run into the embarrassment of being unable to be effectively enforced. Some hold that the clause on “full guaranteed purchase of renewable energy” should be adjusted in light of actual conditions. When the Renewable Energy Law was being drafted, it happened to be when I was in charge of the Energy Administration, and the approving body was the National People’s Congress. During this process, I led teams to report to the NPC many times. The NPC delegates not only listened to my introduction of the Renewable Energy Law but also engaged legal experts to continuously refine the draft of the Renewable Energy Law.

That clause on “full guaranteed purchase of renewable energy” exists in other countries too. Compared with conventional energy, renewable energy is still weak and small, and such a protective policy is highly necessary. Our current proportion of new energy is only one-point-something percent, less than 2%; we are capable of purchasing this small amount of electricity, and I believe there should be no controversy over this. Of China’s 4.95 trillion kilowatt-hours of electricity in 2013, wind power was only 100.4 billion kilowatt-hours, accounting for just 2%.

Full purchase should not have been a problem, so why has it now become a big problem? It’s because of “wind curtailment.” According to data from relevant parties, “wind curtailment” in 2013 was about 20 billion kilowatt-hours, mainly in the Northeast region and Inner Mongolia. The causes of “wind curtailment” include both objective and subjective reasons.

The objective reason is that wind power development in these places is relatively concentrated; although it is not much for the country as a whole, in local areas it might be as high as 20%–30%. In winter the contradiction becomes even more acute, because winter is precisely when the wind is stronger and more wind power can be generated. But at this time heating must be supplied; in northern regions the heat-supply units must first be guaranteed to operate, and if wind and thermal power generation come into conflict and the electricity can’t be consumed, wind power must be shut down to give priority to the heat-supply units. The dispatch order puts heat-supply units in first place, and wind power must give way, so “wind curtailment” arises. But if the whole country could be interconnected and the electricity could be sent out, this problem would no longer exist.

In winter, the people of Inner Mongolia need heating, so the heat-supply units generate, and at the same time wind power can also generate, with the electricity sent to Jiangsu through transmission and transformation lines—then this problem would be solved. But the problem now is that the transmission and transformation lines don’t exist; it can’t be sent out, so on whose head should the blame fall?

On the surface, everyone blames the grid company—you didn’t build the grid. But the grid company doesn’t agree, saying the responsibility lies with the National Energy Administration for being slow to approve the grid plan; without approval, how can it be built? Why no approval? It still comes down to the dispute over ultra-high-voltage (UHV): some favor UHV, some don’t. There are also those who oppose connecting to the East China Grid, saying that if you send it to East China you’d turn the North China Grid and the East China Grid into a synchronous grid; they only agreed to send it to Shandong, which would still be within the North China Grid. But the grid company insisted on sending it all the way to East China, and the dispute went on endlessly with no one to make the call. No one could make the call, and it could only be argued over endlessly. Having disputes is quite normal—which major Chinese project has not had disputes? Didn’t the Three Gorges have disputes? High-speed rail had disputes too, and is still being disputed even now. But should the dispute go on for a long time, endlessly? The transmission and transformation line from Xilinhot to Nanjing still has not been decided to this day. Next, it won’t just be 20 billion kWh of wind power curtailed—there is an even more astonishing figure: in 2014 the Xiluodu and Xiangjiaba hydropower stations will also curtail 20 billion kWh of hydropower. They began generating in 2014, but the transmission and transformation lines have not yet been completed and will lag by one to one and a half years. The main problem with the lag in transmission and transformation line construction is the lag in approval. Half of the five years has already passed, and the most important grid “Twelfth Five-Year Plan” still has not been approved. Why not approved? Because of the constant arguing there. This endless arguing has caused enormous losses to the country. On the one hand we worry about smog weather, while on the other hand we have to give up 40 billion kWh of clean energy. The 100.4 billion kWh is wind power, while the country’s total generation is 4.95 trillion kWh, so wind power’s share is only 2%; therefore consuming this small amount of wind power on a nationwide basis is no problem, but it cannot be consumed in local areas. If there were transmission and transformation lines to send it out, sent to the load centers, this bit of electricity would be no problem at all.

This term of government surely won’t have time to approve it. Whether the next term of government, once it comes in, can approve it immediately, or will wait until the leaders are familiar with it before approving—that’s unknown. But the losses already caused are irreversible. In 2013, clean energy will have to curtail 40 billion kWh—20 billion kWh of wind power plus 20 billion kWh of hydropower. The hydropower is mainly from the Jinsha River, the two lines of Xiluodu and Xiangjiaba, whose completion will be postponed by a year and a half, and some still say don’t make a fuss about this figure. In fact, the situation this year is extremely grave.

So the guaranteed-purchase protective clause for full purchase is not the problem; the problem is that the lag in approval causes renewable energy generation to be unable to be sent to where it is needed. The most fundamental purpose of the Renewable Energy Law is to encourage renewable energy, to create a legal environment for the vigorous development of renewable energy, and to elevate the development of this industry to the legal level for protection. Full purchase is precisely a rigid protective measure that should be strictly observed, but the reality is that we have no way to strictly observe it—not because it can’t be done, but because much of the work has not been done.

When drafting the Renewable Energy Law at the time, we also drew on existing renewable energy policies of countries around the world, among which “full purchase” was also a policy followed in promoting renewable energy in other countries. It is worth noting that the United States has another policy: it stipulates that energy companies have an obligation to develop at least a certain percentage of renewable energy. For example, thermal power enterprises, which in the past burned coal entirely, now have an obligation to add at least 3%–5% new energy.

When the Renewable Energy Law was first formulated, we drew on this clause, and it too was mandatory, requiring power enterprises to have at least an obligation to develop 3%–5% renewable energy. But this clause was later objected to by the legal experts engaged by the NPC when we reported to the NPC; they said that if needed, you could issue an administrative order document to stipulate it, but not write it into the law, so it was removed. At the time it was I who went to the scene to argue the case, but the legal experts simply would not agree, and later no such provision was issued in administrative documents either. So we now have no such mandatory requirement that power enterprises must develop a certain percentage of new energy, and to this day I still feel this is quite a pity.

More to read:

CF40 Research

Why the OECD’s Subsidy-Centric Narrative of China’s Emerging Industries Is Increasingly Flawed and Outdated

Abstract: In a recent report, the OECD contends that the rise of China’s emerging industries has rested largely on government subsidies—with lending from the state-owned banking system as a principal channel—and that these subsidies, more than any other factor, account for the international competitiveness of Chinese firms. We examine this contention using the 2018–2025 annual reports of more than 5,300 non-financial companies listed on China’s A-share market. Classifying firms by R&D intensity and profitability, we divide the sample into a “new economy” and an “old economy”; the former corresponds closely to what the OECD designates as China’s emerging industries, namely electric vehicles, batteries and solar, together with computing, semiconductors and related sectors…

Read more

2 months ago · 1 like · CF40 Research

But , a long-term researcher of the Chinese new energy sector pointed out some of the framing and epistemic problems of the article. He argued that the “turmoil” is better understood as a structured transition. When policy protection was pulled out, and the industry was exposed to the real market, such friction is predictable. Or, to put it in our Chinese words, it’s 发展中出现的问题, and it has to be 靠发展解决—problems that emerge in the course of development, which can only be solved by further development.

In his view, the piece oversimplifies why domestic demand is softening, treating it as a story about grids becoming overloaded with solar when the real driver is the shift away from generous feed-in tariffs toward market-based pricing.

He also believes the article gets the cause and effect on curtailment backwards, describing it as a problem of too much solar when it is really a problem of insufficient grid flexibility. He thinks that the piece mischaracterizes the coal-versus-renewables dynamic, which has less to do with one fuel locking out another and more to do with the basic tension between forward contracting and real-time markets.

Finally, he questions how the article reads both the export picture and the wave of company bankruptcies, suggesting that if oversupply is truly the core problem, then firms exiting the market should count as part of the solution rather than a sign of gloom.

In the end, he argues, the article never lands on a coherent thesis tying its negative data points together.

I decided to share his correction of the article as a sort of contribution to understanding the Chinese new energy sector.

David Fishman currently works for the Lantau Group, where he specializes in China’s power sector. His work covers solar, wind, coal, nuclear, hydro, transmission, and power markets, with particular focus on renewable energy policies and market forecasting. Before joining Lantau Group, Fishman was Managing Co-Director of Shanghai-based Nicobar Group, a nuclear energy consultancy. He holds a joint MA in International Relations and Energy Policy from Johns Hopkins SAIS and Nanjing University, and is fluent in Mandarin Chinese.

He’s also the founder of a newsletter Crossing the River by Feeling the Stones where he provides first-hand observations on Chinese rural areas and poverty alleviation efforts. His conversations with locals offer valuable grassroots perspectives on China’s development, which I really enjoyed.

Thanks to Mr. Fishman for allowing me to repost his counterargument in my newsletter. Below is his piece, as published on LinkedIn.

This is a Structured Transition, Not Turmoil: Responding to The Economist’s Coverage of China’s Solar Sector

Last week The Economist ran this piece (link paywalled) titled “China’s world-beating solar industry is in turmoil“:

While it does cover some important and correct aspects of the current challenges in China’s solar sector, including low panel prices and a slowdown in capacity growth, I also found a number of problems with the piece - both epistemic and in framing/presentation.

I then became increasingly concerned to see the piece start to be quoted in some of my industry circles. This prompted me to prepare a response for Twitter/X, which I will now reproduce here on LinkedIn as well.

For each section, I will provide a screenshot of text from The Economist article, with relevant sections highlighted in yellow and my commentary following the screenshot. Let’s start:

1) Oversimplification

Unfortunately, it is a major oversimplification to say Chinese domestic demand for solar panels is falling “because the country’s power grids have become overloaded with the things”.

The real challenge right now is developers and banks still figuring out how to finance and build projects without policy-backed revenue guarantees. Domestic demand was shored up in past years by a generous feed-in-tariff (FiT) scheme that ensured stable and predictable revenues. Following the the longstanding policy trend toward liberalization, last year solar generators were shifted out of FiTs and into a Contract-for-Difference (CfD) scheme. Guaranteed CfD volumes have quotas, with non-CfD volumes expected to find customers via open markets (i.e. merchant exposure). The piece acknowledges the policy change from last year, but doesn’t seem to recognize the importance.

New solar capacity growth will likely be flat or even decline YoY in 2026 because generators and financiers are still inexperienced with merchant risk and renewable consumption quotas aren’t quite high enough yet to drive more long-term renewable power contracting demand (which financiers rely on).

Yes, low prices in daytime spot power markets reflect temporal oversupply, but that’s largely irrelevant for investment decisions, which are built around long-term contracts, not spot markets.

2) Misdiagnosis

The cause-and-effect relationships in this paragraph are backwards.

If you’re explicitly trying to build a low-carbon grid, solar curtailment is better seen as a problem of insufficient flexibility (via storage or conventional generators), not “too much solar.”

Conventional coal typically can’t ramp up or down quickly...in many cases it takes hours, or sometimes even the better part of a day for older plants. It is the lack of flexibility that leads to curtailment of assets you ideally don’t want to curtail...like renewables. When solar performance peaks at noon, you need other assets to be able to rapidly ramp down to make room. Not incidentally, the new Chinese coal plants being built these days are capable of flexible operations, and older plants are being retrofit to do the same.

Finally, there are no “shortages at night” because of solar only working during the day. System operators plan for the sun not shining...

3) Mischaracterization

I don’t agree with the characterization that “long-term contracts for coal-fired power lock out renewables even though they are cheaper”. It’s more complex and nuanced than that.

Long-term contracts for anything can lock out anything not contracted. This is less of a question of “coal vs renewables” and more “forward contracting vs real-time markets”.

This is a fundamental friction in power markets. Electricity is contracted ahead of time but dispatched and consumed in real time. You forecast load as best you can when you contract, but when it comes to dispatch, you have to match actual load with actual generation...and forward-contracted generation typically runs first. If total contracted generation exceeds load, then the system must deviate from contracts, typically by allocating curtailment proportionally.

I’d like to see the Chinese power market move away from such high reliance on annual contracting and more use of spot market price signals - and it’s going in that direction. But this can also highlight another problem: renewables often need long-term contracts to get assets financed in the first place and would never want to be reliant on spot market offtake.

Whatever workable balance China ends up with, it’s going to be achieved slowly, via trial and error.

Finally, given the current buildout of storage, UHV, and flexible generation, it’s not clear why solar installs couldn’t re-accelerate in 2027-2028. Renewable consumption quotas are only rising each year, including for data centers, so demand is only going to keep rising too.

4) Misattribution of Intent

While these details are broadly correct, I wouldn’t frame this as evidence that officials are trying to curb a bloat in solar projects. Renewable subsidies and guarantees have been gradually stripped away over the last decade, well before current market conditions.

The big policy change last year (the infamous Document 136) that shifted projects out of the guaranteed FiT scheme and into CfDs or fully market-based compensation is a continuation of long-running market liberalization trends.

Local officials aren’t anti-solar per se; they just don’t want to see rising curtailment rates (one of their KPIs). They are increasingly incentivized towards approving new projects with clearly identified offtake.

5) Misstatement of Impact

The US hasn’t been a major direct destination of Chinese solar exports since the early 2010s. The major export market has historically been the EU, not the US.

Inverters are not solar panels and the EU remains a big export market for Chinese solar.

6) Missing the Solution For the Problem

By the logic of this article so far, 40 solar companies going under should be cause for optimism, not a sign of “gloom”. If oversupply is the driving problem, then consolidation of the sector via producer exit is part of the solution, no?

This section also starts to mix up two groups with opposing incentives: Solar panel producers and solar panel buyers (i.e., developers). The former wants higher prices; the latter wants lower prices. Anything that benefits one side on price is going to hurt the other group. These two segments should not be collapsed into the same commentary.

Expanding back into blocked export markets likely doesn’t fix this, even if they were big markets (and considering the successful efforts to cultivate new export markets and triangulate Chinese products via third-country production sites, it’s not even clear how meaningful the blocks have been).

Also, developing export markets are no guarantee of profitability. For solar OEMs, oversupply brings the same vicious price competition in overseas markets. Higher volumes don’t help that much if margins are still extremely thin - and they hurt if margins are negative.

In Closing:

Stepping back, the biggest issue with the piece is that it never really lands on a thesis. It puts together a set of negative data points like falling prices, curtailment, project delays, and pressure on firms, but never explains what ties them together or what it all means.

Most crucially, it never acknowledges that most of these tensions aren’t even particularly weird. They’re the basic mechanics of power markets integrating renewables. Electricity is contracted ahead of time but delivered in real time. Low marginal cost generation smashes prices during peak performance hours. Flexibility always lags capacity. Financing depends on long-term certainty that markets don’t naturally provide. If you don’t know how to anchor the analysis in those facts, I suppose pretty much everything starts to look like chaotic dysfunction.

What’s actually happening is much simpler. China is scaling solar to a level where abundance, not scarcity, becomes an actual binding constraint. The whole sector is getting pulled out of policy protection and into market exposure. Of course that produces friction. It’s supposed to.

You can call it “turmoil” if you want. But this is what it looks like when a sector stops being policy-supported and starts behaving like a real market system.

Thanks for reading.

After returning to China in 2013, Miao served at the State Administration of Foreign Exchange (SAFE) for a decade. From 2013 to 2018, he served as Senior Advisor to the Administrator of SAFE and as Head of Research at the SAFE Investment Center, where he led global macro and strategy research related to the management of foreign exchange reserves. In May 2018, he was appointed Chief Economist of the SAFE Investment Center. In 2023, Miao joined CICC. In July 2024, he attended an expert symposium on the economic situation chaired by Premier Li Qiang.

The article takes the United States’ “countercyclical opening” of its capital account in 1974 as a case study, revisiting how—at a moment of damaged credibility and stagflation following the collapse of the Bretton Woods system—the U.S. managed to rebuild the “dollar recycling” mechanism and consolidate the dollar’s status as the global reserve currency precisely by opening up. But I believe the author also intends to offer policy recommendations for RMB internationalization.

Miao argues that China should move beyond the static mindset that equates opening up with increased risk exposure and instead adopt a dynamic perspective on trade-offs. He believes that, in the short term, capital account liberalization may indeed bring a period of pain brought by capital outflows and exchange rate volatility, similar to what the United States experienced between 1974 and 1982. Yet so long as a sustainable “outflow–inflow” cycle can be established through institutional arrangements such as greater exchange rate flexibility, the coordinated development of onshore and offshore markets, and the expansion of RMB-denominated invoicing and settlement networks, opening up will, through the very dynamics of two-way capital flows, drive market pricing mechanisms, hedging instruments, and the macroprudential framework toward greater maturity, which ultimately strengthening the resilience of China’s financial system against external shocks. Prolonged closure, the author contends, carries its own costs: if RMB-denominated assets cannot be conveniently held and allocated by overseas entities, the RMB will struggle to evolve from a settlement currency into an investment and reserve currency.

The Chinese version can be found on the webpage of the CF40 Forum. Thanks to Dr.Miao’s kind authorization, I can publish the English version in my newsletter.

How the United States Opened Its Capital Account Against the Tide

The United States today is often viewed as a symbol of free capital flows and financial openness. Yet what is little known is that the complete opening of its capital account did not happen overnight. Rather, it is a history of dynamic trade-offs centered on the interests of reserve-currency status, balancing the benefits of dollar hegemony against the risks to financial stability.

From the late 19th century to the eve of World War I, the United States was broadly part of an open system under the gold standard, but international financial centers were still dominated by London, and the dollar’s international financial functions were limited. It was only after the establishment of the Federal Reserve in 1913 that the United States, by institutionalizing the Banker’s Acceptance market and providing discount support, propelled the dollar toward internationalization.

During the Great Depression and World War II, in order to maintain financial stability and the security of gold reserves, the United States imposed controls on gold and capital flows. After World War II, the Bretton Woods system established the dollar’s status as the central currency, but it also saddled the dollar with the rigid commitment of being pegged to gold.

In the 1960s, as the U.S. balance of payments deteriorated and pressure from gold outflows intensified, the United States—in order to maintain fixed exchange rates and the dollar’s reserve-currency status—was forced to introduce a series of capital control measures such as the Interest Equalization Tax and restrictions on foreign credit.

However, entering the 1970s, capital account controls did not truly prevent the arrival of the dollar crisis; instead, they gradually came to threaten the dollar’s reserve-currency status. As global trade and cross-border capital expanded rapidly, the irreconcilable contradiction between fixed exchange rates and cross-border capital flows continued to intensify. In 1971, the Nixon administration was forced to announce the closing of the gold window, the Bretton Woods system moved toward collapse, the dollar depreciated sharply against major currencies, and the dollar’s credibility suffered a severe blow.

At the same time, the increasingly powerful offshore dollar market gradually weakened the effectiveness of U.S. onshore controls, and after the oil crisis, the massive dollar surpluses of oil-producing countries also needed a sufficiently open and deep financial market to absorb them. At this point, if the United States clung to capital controls, it might not truly prevent dollar outflows, but could instead weaken the position of the New York financial center and forfeit the reserve-currency dividend.

To consolidate the dollar’s reserve-currency status, amid the adversity of shaken dollar credibility, high global inflation, and a U.S. economy mired in stagflation and recession, in 1974 the United States instead chose to loosen restrictions on capital flows and promote the opening of the capital account.

In hindsight, this “opening against the tide” was indeed accompanied by net capital outflows and dollar depreciation in the short term, but over the long run it became the starting point for the dollar system to move from crisis toward strengthening. In the decade that followed, the United States absorbed global surplus funds through petrodollar recycling, restored price stability and reshaped dollar credibility during the Volcker era, and unleashed growth momentum with the supply-side economics of the Reagan administration—gradually forming a dollar cycle that allowed the dollar system to complete a restructuring from crisis to reinforcement.

So why did loosening capital controls at a moment of dollar fragility ultimately not weaken the dollar system, but instead become the starting point for its renewed strengthening? Was this choice a proactive plan by the United States to reshape the dollar cycle, or a passive adjustment forced by crisis?

More importantly, today—when dollar credibility is once again being questioned and the global monetary system is fragmenting at an accelerating pace—what lessons does the U.S. experience hold for the current internationalization of the renminbi and for high-level capital-account opening? This article will analyze the questions above.

I. How to Open? Under the Twin Crises of Dollar Credibility and the U.S. Economy, Rebuilding the “Dollar Cycle” Through Openness to Consolidate Reserve-Currency Status

In the early 1970s, the United States faced a critical turning point for dollar credibility following the collapse of the Bretton Woods system. In August 1971, with international capital flowing at an accelerating pace and the dollar–gold convertibility commitment becoming unsustainable, then-President Nixon announced the closing of the gold window, decoupling the dollar from gold and leaving the Bretton Woods system in name only.

In less than a year that followed, the dollar depreciated markedly against the currencies of major developed economies, and the dollar’s share of global reserves—which had reached a peak of 85% in the fourth quarter of 1970—fell rapidly. In March 1973, the dollar depreciated again, but this failed to stabilize market confidence; a large-scale dollar sell-off emerged in international foreign exchange markets, major European currencies were forced to shift to floating exchange rates, and the Bretton Woods system collapsed completely.

Just as dollar credibility fell into difficulty, the first oil crisis further amplified U.S. macroeconomic pressures—yet it also unexpectedly opened a window for rebuilding the dollar cycle.

In October 1973, the Fourth Middle East War broke out, OPEC announced an oil embargo, and the first oil crisis erupted in full. The rapid rise in oil prices, on the one hand, brought pressure to the U.S. economy—the U.S. economy fell into stagflation and ultimately entered recession in the fourth quarter of 1973. On the other hand, it also caused OPEC countries to accumulate large dollar surpluses, while U.S. financial markets—especially the Treasury market—possessed asset capacity and liquidity advantages rare in the world.

At the same time, the rise in oil prices also created demand for dollar payments. Combined with the fact that the oil crisis hit economies heavily dependent on Middle Eastern oil—such as Europe and Japan—even harder, these factors together pushed the dollar index from 94 before the first oil crisis to 109 in January 1974, briefly “steadying” a dollar that had been depreciating successively since the collapse of Bretton Woods.

At this point the United States faced a dilemma over whether to open its capital account: the United States urgently needed Middle Eastern petrodollar funds to flow back into the Treasury market, in order to consolidate demand for dollar assets and provide financing support for a U.S. economy in recession. At the same time, as cross-border capital flows became increasingly frequent, exchange rates turned to floating, and the offshore dollar market developed, capital controls became increasingly difficult to maintain (Hawley, 1987).

But the problem was that opening the capital account amid economic recession and a shaken dollar position could also amplify pressures from capital outflows and dollar depreciation. Facing this trade-off, Paul Volcker wrote in his memoirs that, in the context of 1974, opening the capital account to allow foreign funds into New York’s capital markets was more advantageous (Volcker & Gyohten, 1992).

In January 1974, the United States chose to open its capital account amid economic recession. Specific measures included:

1. Abolishing the Interest Equalization Tax (IET) established in the 1960s to prevent capital outflows, so that U.S. investors purchasing foreign securities would no longer be taxed.

2. Abolishing Balance of Payments Controls and loosening restrictions on outward foreign direct investment (OFDI) by U.S. residents.

3. Abolishing the Voluntary Foreign Credit Restraint (VFCR) and other limits on the amount of overseas lending by banks.

Subsequently, in the 1980s the United States gradually relaxed “Regulation Q,” which capped bank deposit interest rates, and formally abolished the relevant arrangements in 1986, while also establishing International Banking Facilities (IBFs)—moving toward financial deregulation and further embracing offshore markets.

In the early stage of capital-account opening, the United States did indeed experience a “painful period” of economic recession, capital outflows, and dollar depreciation. After opening the capital account in January 1974, the U.S. economy remained in recession. It was not until March 1975 that it bottomed out and stabilized.

In the early stage of opening the capital account, the United States did indeed see net capital outflows. Although in mid-1974 the United States signed a petrodollar agreement with Saudi Arabia, stipulating that Saudi Arabia would invest part of its oil revenue in Treasuries—boosting inflows under the portfolio investment item of the U.S. financial account—the simultaneous expansion of overseas lending by U.S. banks led to outflows under the other investment item.

The result was that from 1974 to 1982, the U.S. financial account overall still showed net outflows abroad, briefly turning to net inflows only from 1977 to 1979. The economic downturn and capital outflows reinforced each other, pushing the dollar index down from its peak of 109 in January 1974 to a low of 82 in October 1978—a 25% depreciation in less than five years.

But over a longer horizon, opening the capital account did not weaken the dollar system; instead, it became a necessary condition for rebuilding the “dollar cycle” and re-steadying dollar hegemony. The key was not whether capital flowed back immediately after opening, but that the United States provided global dollar funds with a financial market that could be entered, settled into, and allocated.

It was precisely under the combined effects of petrodollar reflux after capital-account opening, the repair of dollar credibility, and steady U.S. growth that the United States, from the late 1970s through the 1980s, gradually completed the restructuring of the dollar system from crisis-induced pressure to a strengthened cycle.

► The petrodollar mechanism promoted dollar reflux: In July 1974, then-Treasury Secretary William E. Simon traveled to Saudi Arabia and reached a secret agreement. Although the dollar had already dominated the pricing of international oil trade before the petrodollar mechanism formally took effect, the agreement further consolidated and institutionalized the dollar’s dominant position as the currency for pricing and settling oil.

More importantly, Saudi Arabia committed to investing part of its dollar revenue from oil exports in U.S. Treasuries, thereby forming an important reflux mechanism for the dollar cycle. In exchange, the United States provided military protection to Saudi Arabia, and the U.S. Treasury provided Saudi Arabia with an “Add-ons” special window for Treasuries, allowing Saudi Arabia to purchase Treasuries directly, bypassing open competitive bidding.

► In the 1980s, Volcker restored price stability and reshaped dollar credibility, while the Reagan administration’s supply-side economics stimulated growth, forming a positive feedback loop: In 1979, Paul Volcker became Chairman of the Federal Reserve, placing inflation control at the core of policy and stabilizing prices through tough tightening—even at the risk of triggering recession—to revive market confidence in the dollar. The federal funds rate at one point rose above 20% in the early 1980s; high interest rates suppressed U.S. inflation and rebuilt the dollar’s purchasing power, with the year-over-year U.S. CPI falling from 14.8% in March 1980 to 2.5% in July 1983.

The process of monetary tightening came at a high cost: the United States fell into recession twice, in February 1980 and August 1981, but it also raised the real returns on dollar assets, attracting global funds into U.S. bond and deposit markets and laying the foundation for rebuilding dollar credibility.

In 1981, Ronald Reagan became President of the United States, and facing economic recession he adopted supply-side propositions to stimulate growth, spurring investment and supply recovery through tax cuts, accelerated depreciation, deregulation, and other measures. Although the goal of “small government” was not truly achieved—the U.S. fiscal deficit ratio (monthly average of -4.1%) actually expanded compared with the 1970s (monthly average of -1.9%)—fiscal expansion, tax incentives, and financial liberalization together boosted U.S. economic growth.

U.S. real GDP growth, after touching a low of -2.6% in the third quarter of 1982, entered a nearly two-year upward channel, reaching a peak of 8.6% in the first quarter of 1984. The dollar also rose sharply from its low of 85 in 1980 to a peak of 165 before the 1985 Plaza Accord.

As a result, the United States formed a self-reinforcing “dollar cycle,” allowing the dollar to strengthen against the tide despite the U.S. twin fiscal and trade deficits. Price stability enhanced dollar credibility, high interest rates raised returns on dollar assets, growth recovery enhanced the appeal of U.S. assets, and capital inflows in turn provided financing support for U.S. fiscal expansion and financial-market prosperity. At the same time, the strong domestic demand brought by economic growth and a strong dollar expanded U.S. import demand, pushing up the trade deficit. And the dollar’s special status as an international reserve currency meant that the dollars flowing out through imports returned in the form of financial investment.

This mutually reinforcing process was summarized by Soros in The Alchemy of Finance as “Reagan’s Imperial Circle.” After 1983, the U.S. financial account turned to net inflows, which have continued to this day.

II. Why Open? Proactive Restructuring Under Passive Pressure—The Core Being That Capital Controls Harm Reserve-Currency Status in the Long Run

Taken together, the U.S. opening of its capital account in the 1970s, at a “fragile moment” of economic recession and dollar-credibility pressure, was essentially a proactive choice under passive shock.

It was “passive” because, after the collapse of the Bretton Woods system, the fixed-exchange-rate objective that capital controls were originally meant to serve had already become invalid; at the same time, the rapid expansion of the offshore dollar market made domestic capital controls increasingly difficult to enforce.

It was “proactive” because the United States realized that continuing to restrict capital outflows might not stop dollars from flowing out, but could instead weaken the central position of dollar assets and the New York financial market, ultimately damaging dollar credibility. Rather than clinging to an already-invalid closed framework, it was better to rebuild, through openness, the cycle of dollar funds “flowing out—flowing back.”

It is precisely in this sense that the U.S. capital-account opening in the 1970s was both a result of crisis-driven pressure and the starting point for a proactive restructuring of the dollar system.

Passive shock: The floating exchange-rate system weakened the necessity of capital controls, and the expansion of the offshore dollar system forced capital-account opening

First, under the floating exchange-rate system after the collapse of Bretton Woods, the necessity of implementing capital controls declined. According to Mundell’s “impossible trinity,” a country cannot simultaneously achieve all three goals of fully independent monetary policy, exchange-rate stability, and free capital flows (Mundell, 1963; Obstfeld and Taylor, 1997). During the Bretton Woods period, countries in effect restricted capital flows as the price for obtaining fixed exchange rates and a degree of policy autonomy.

The U.S. restrictions on residents purchasing foreign securities, on bank overseas lending, and on corporate overseas investment were all aimed at reducing dollar outflows and gold-convertibility pressure. Volcker recalled that the U.S. capital controls of the 1960s came about when continuous capital outflows pressured the dollar, leaving the Treasury no choice but to use tools such as the Interest Equalization Tax (Volcker & Gyohten, 1992). However, the irreconcilable tension between fixed exchange rates and cross-border capital flows ultimately still dissolved the Bretton Woods system (Eichengreen, 1996; Obstfeld and Taylor, 1997); the fixed-exchange-rate system thereupon exited the stage, the institutional benefits of continuing capital controls declined, and the policy costs rose.

Second, the development of the offshore dollar market also forced the United States to loosen capital controls. The offshore dollar market (also known as the Eurodollar market) originated in the 1950s. After World War II, the United States continuously exported dollars to Europe through the Marshall Plan, overseas military spending, and overseas investment. Against the backdrop of the Cold War, countries such as the Soviet Union, in order to evade U.S. regulation, deposited dollars in the London banking system, and the earliest offshore dollar deposits were thus formed.

Entering the 1960s, financial regulatory measures within the United States further enhanced the appeal of the offshore dollar market. At the time, banks within the United States were constrained by “Regulation Q,” with caps on deposit interest rates, and also had to bear regulatory costs such as reserve requirements, whereas offshore dollar deposits were not subject to such constraints. For the same dollar liabilities, onshore costs were higher and constraints more numerous, while offshore costs were lower and restrictions fewer, giving funds and business an incentive to migrate to offshore markets (Friedman, 1971).

Against this backdrop, the offshore dollar market expanded. Friedman vividly likened the formation and expansion of the offshore dollar market to “the bookkeeper’s pen,” emphasizing that offshore dollars derived neither primarily from the U.S. balance-of-payments deficit or various central banks’ reserves, nor from the large-scale cross-border transfer of dollar cash, but were the result of credit creation by the banking system (Friedman, 1971). U.S. capital controls were facing a system that could continuously derive dollar credit beyond regulatory boundaries, rather than simple capital outflows.

However, Friedman did not believe the offshore dollar market could expand infinitely; he used “leakages” to explain the boundary of its credit derivation (Friedman, 1971). So-called leakage refers to dollars lent out by offshore banks that do not return to the Eurodollar banking system, but instead settle into deposits at U.S. domestic banks, are used for foreign exchange conversion or trade payments, or go into other asset allocation.

If a borrower, after obtaining dollars, deposits them in another offshore dollar bank, the new bank can continue lending after retaining a small amount of prudential reserves, and the offshore dollar multiplier rises; conversely, if the borrower uses the funds for payments, currency conversion, or deposits them in a non-Eurodollar bank (for example, depositing them back into a U.S. domestic bank), that portion of funds is equivalent to “leaking out” of the Eurodollar system, the credit chain is broken, and the derivation multiple declines accordingly.

The lower the leakage, the higher the offshore re-deposit ratio, and the more easily dollar credit can be derived layer by layer overseas; the higher the leakage, the closer offshore activity comes to a one-off bookkeeping transfer, and the weaker the multiplier effect. Therefore, regulatory arbitrage opened up the space for offshore dollar credit creation, while leakage determined the extent to which derived credit could be magnified.

Even with the constraint of leakage, the offshore dollar market had already reached a considerable scale by the early 1970s. According to BIS statistics, by the end of 1973 the external dollar liabilities of reporting European banks had reached 130 billion dollars and external dollar assets nearly 134 billion dollars, having grown roughly tenfold in just seven years, with an average annual expansion rate of 20–30%—far exceeding the growth rate of the U.S. domestic money supply over the same period.

Even after deducting the double-counting caused by re-deposits within the European banking system, the dollar credit scale estimated by the BIS for 1973 was still about 97 billion dollars. This scale was equivalent to 7% of U.S. GDP in 1973, 1.4 times the full-year merchandise imports, and 53% of total global official reserves—already enough to affect dollar funding costs, cross-border credit allocation, and the competitiveness of the New York financial center.

Thus, the growth of the offshore dollar market created pressure compelling the United States to open its capital account. On the one hand, the rapid expansion of the offshore dollar market rendered capital controls “existent in name only.” Global demand for dollars did not weaken because the United States tightened onshore controls; instead, it shifted more toward the offshore dollar market.

In a 1971 memorandum, Shultz pointed out that the effectiveness of capital control measures was questionable: the Interest Equalization Tax reduced foreign securities purchases, but this gap was quickly diluted by increases through channels such as U.S. direct investment and bank loans; and although foreign credit restraints once curbed short-term private U.S. capital outflows, this positive effect was almost entirely offset by reverse changes in foreign private short-term capital flows. Therefore, capital controls did not truly improve the balance of payments, and even brought economic-efficiency losses and administrative costs.

On the other hand, this also put pressure on the pricing system of the onshore dollar financial system. The offshore market could conduct dollar business at lower cost, gradually forming an offshore interest-rate system represented by LIBOR. Volcker also pointed out that if the United States continued to insist on capital controls, demand for dollar funding and trading would still exist; global dollar business would not disappear, but would simply shift more to offshore centers such as London, the Bahamas, and Luxembourg, harming New York’s intermediary and pricing position in global dollar funding (Volcker & Gyohten, 1992).

Proactive choice: Strengthening the dollar’s reserve-currency status; the neoliberal current of thought provided the policy atmosphere; an opportunity to re-steady the dollar system

Beyond the passive adjustment driven by external shocks, the U.S. choice to open its capital account in 1974 also had its proactive strategic considerations. First, the foundation of openness lay in the dollar’s reserve-currency status. After the collapse of the Bretton Woods system, the dollar’s share of reserve currencies declined somewhat, and although there was still no currency that could replace the dollar (There is no alternative, TINA), concerns persisted that long-term capital controls would weaken the usability and credibility of dollar assets.

Volcker recalled that Shultz and others had real reasons for opposing capital controls: the effectiveness of controls was increasingly limited, the distortions increasingly large, enforcement increasingly difficult, and they had become a burden on businesses and financial institutions (Volcker & Gyohten, 1992).

In addition, the “neoliberal” current of thought in the 1970s also provided the policy atmosphere for the United States to open its capital account. The stagflation shock of the 1970s weakened the Keynesian policy consensus, “neoliberalism”—with monetarism and free markets as its main ideas—rose, and financial liberalization gradually became the mainstream policy direction (Harvey, 2005).

Volcker recalled that officials such as Shultz were deeply influenced by Friedman and believed that markets could allocate capital better than governments (Volcker & Harper, 2018). Simon, too, repeatedly emphasized in his memoirs his opposition to government control and his advocacy of a return to free-enterprise principles (Simon & Caher, 2003).

Against this backdrop, the United States realized that opening the capital account was not merely a source of risk, but could also become an opportunity to re-steady the dollar system. As the issuer of the reserve currency, although lifting capital restrictions might cause dollar outflows, paths such as global trade, reserves, and safe-asset allocation were still creating demand for dollars. Therefore, rather than clinging to invalid capital controls and watching the offshore market erode the influence of the New York financial center, it was better to open proactively, leveraging the depth and liquidity of U.S. financial markets to re-absorb global funds—including petrodollars—back into the domestic system, consolidate the position of the New York financial center, and rebuild the dollar cycle to steady the dollar’s hegemonic position. Dollars would flow out, but they would also flow back through global dollar demand, the allocation of oil-producer surplus funds, and the depth of U.S. financial markets.

Smooth operation after opening: The depth of financial markets and technological/institutional advantages were important supports

After the United States opened its capital account in 1974, no stable net foreign inflows formed in the initial period, and it even bore net capital-outflow pressure for a time. But these outflows did not evolve into uncontrolled capital flight, and the dollar system did not collapse as a result. The reason was that behind it lay financial markets deep enough to absorb the funds, relatively mature payment-clearing and trading infrastructure to support liquidity, and relatively complete institutional arrangements to stabilize expectations. This was specifically reflected in the following three aspects:

► Market depth: The Treasury market was large in scale and strong in liquidity, serving as an important safe-asset pool for global capital. By the 1970s the United States had gradually formed relatively mature mechanisms for Treasury issuance, pricing, and trading; medium- and long-term Treasuries began to be auctioned regularly, and short-term Treasury Bills were continuously rolled over.

At the same time, multi-tiered short-term money markets—including the federal funds market, large negotiable certificates of deposit (CDs), and the repo market—also developed rapidly, reducing friction costs for the banking system’s liquidity allocation (Gorton, 2010).

► Technological advantages: In the 1970s the technology supporting cross-border capital flows became increasingly mature; the Clearing House Interbank Payments System (CHIPS) and the Society for Worldwide Interbank Financial Telecommunication (SWIFT) were established in 1970 and 1973, respectively. Among them, CHIPS served large-value net dollar payments and was widely used for settlement related to foreign exchange and securities trading, serving as the core clearing infrastructure supporting cross-border dollar transactions; SWIFT promoted the standardization of cross-border financial communication, facilitating cross-border capital flows.

► Institutional completeness: Financial regulatory systems were mature; domestically, the securities market had established a regulatory system under the U.S. Securities and Exchange Commission (SEC), the banking system had deposit insurance provided by the Federal Deposit Insurance Corporation (FDIC), and the Federal Reserve assumed the lender-of-last-resort function. Externally, the Treasury maintained dollar stability through the U.S. Treasury’s Exchange Stabilization Fund (ESF), while the Federal Reserve established liquidity-support mechanisms with major central banks through central-bank swap lines—together constituting the institutional support for the dollar’s external stability.

Volcker mentioned in his memoirs that Robert Roosa called these official financial innovations the “inner and outer lines of defense” for protecting the dollar and the international monetary system (Volcker & Gyohten, 1992); the existence of these arrangements shows that even as the United States loosened capital controls, it did not abandon management of the dollar system.

From this it can be seen that capital-account opening did not occur after “everything was ready”; it was more a trade-off made when old constraints were gradually loosening and a new cycle had not yet formed. Other countries had similar experiences. In October 1979 the United Kingdom lifted its remaining foreign-exchange controls, facing the decline of sterling’s status and pressure to reposition the London financial center; had it continued controls, it might not have held onto sterling, but could have lost its financial center. In 1979 Japan revised its Foreign Exchange and Foreign Trade Control Law, shifting from “prohibition in principle” to “freedom in principle”; West Germany in 1981 basically removed remaining restrictions on capital inflows and in 1984 further abolished the withholding tax on non-residents holding German bonds—facing current-account surpluses, currency appreciation, and demand for capital export; had they continued to remain closed, the external surpluses accumulated by the real sector would have been hard to convert into international financial influence.

Therefore, the key to capital-account opening lies not in waiting for a moment of complete stability and the right internal and external environment, but in whether, under controllable risk, passive shocks can be transformed into proactive restructuring, and whether dynamic adjustments can be continuously made during the opening process—to magnify the benefits of opening and shrink the risks of opening.

III. Lessons for Renminbi Internationalization? Establishing a Mindset of “Dynamic Trade-offs” in Opening; Capital-Account Opening and Currency Internationalization Dynamically Reinforce Each Other

At present, China stands at a critical historical juncture for steadily expanding institutional opening and promoting renminbi internationalization during the “15th Five-Year Plan.” According to the IMF’s classification of capital-account transactions, China has achieved varying degrees of convertibility for most items, but a few sub-items still implement relatively prudent management; compared with major reserve-currency economies, there is still room to raise the actual degree of financial openness. Concerning capital-account opening, the most common market concern is: will further raising the level of capital-account openness lead to large-scale capital outflows?

Referring to the U.S. experience, the early stage of capital-account opening may indeed bring the pain of capital outflows. But the crux of the matter lies not in whether capital outflows occur in the early stage of opening, but in whether a sustainable funding cycle can form after opening—and this ultimately depends on whether prices are stable and growth is robust.

In the early stage of currency internationalization, capital-account opening is not a precondition; but in the long run, an internationalized currency must be freely convertible. As a reserve currency, it must open its capital account, exporting liquidity overseas, while overseas demand in turn prompts the currency to flow back, forming a cycle of two-way flows.

The history of U.S. capital-account opening is, in essence, a history of dynamic trade-offs between strengthening reserve-currency status and maintaining financial stability. In this process of dynamic balance, the reserve-currency dividend and the financial-stability risk brought by opening are not fixed. In the long run, once the internationalization network and the reflux mechanism mature, the institutional benefits brought by opening will continue to magnify, and the financial system’s resilience to risk will also continuously strengthen amid the two-way cycle.

The U.S. case shows that the reason capital-account opening was ultimately able to strengthen the dollar system lies not in opening itself, but in the institutional synergy formed among opening, floating exchange rates, offshore-market development, financial-market depth, and macroeconomic stability.

First, long-term capital controls weaken the availability and credibility of reserve-currency assets, ultimately disadvantaging reserve-currency status. Second, floating exchange rates and offshore-market development are important conditions for the capital account to be able to withstand the shocks of two-way flows. Third, whether funds can continuously flow back after opening depends on whether the home country’s assets are sufficiently attractive—which in turn depends on price stability, growth resilience, and financial-market depth.

Thus, the lessons for renminbi internationalization are as follows:

► Establish a mindset of “dynamic trade-offs” in opening, driving a “Pareto improvement” in macro-finance through high-level opening to the outside world. For the United States in 1974, between the risk of capital outflows and the benefit of consolidating reserve-currency status, it chose to open its capital account. Although it faced the pain of capital outflows and exchange-rate depreciation in the early stage, it ultimately still became the starting point for strengthening the dollar’s status.

For China, “no sugarcane is sweet at both ends”—there is no completely risk-free shortcut to capital-account opening, and the process of opening must be a dynamic trade-off between risk and benefit. Statically, relaxing capital controls a bit more seems to mean exposing a bit more outflow risk. However, dynamically, long-term closure also comes at a cost. If renminbi assets are difficult for overseas entities to conveniently hold and allocate, the renminbi will struggle to move further from a settlement currency to an investment currency and a reserve currency, and thus will be unable to obtain the reserve-currency dividend.

Therefore, the significance of high-level institutional opening is not simply to enlarge risk exposure, but to reshape the risk-return combination at a higher level, achieving a “Pareto improvement” in the macro-financial sense: enhancing both the efficiency of resource allocation and the appeal of renminbi assets, while also—through more mature market pricing, richer risk-hedging tools, and a more complete macroprudential framework—improving the entire economy’s ability to withstand external shocks. In other words, opening is not about sacrificing safety for efficiency, but about simultaneously enhancing both efficiency and safety at a higher level.

Just as a person can only learn to swim in the water, the renminbi must also learn to open within openness; it cannot learn to open within closure. Actively advancing high-level institutional opening, in an environment of two-way capital flows, helps market participants better develop pricing, hedging, and risk-management capabilities, and only thus can the macro-policy framework be continuously improved. From this, the renminbi can achieve a dynamic unity of high-level development and high-level security, and the process of dynamic capital-account opening will also mutually reinforce the historical process of renminbi internationalization.

► Increase exchange-rate flexibility, and coordinate the development of onshore and offshore renminbi markets. An important precondition for the U.S. opening of its capital account in 1974 was the floating exchange-rate system after the collapse of the Bretton Woods system; exchange-rate floating is the “safety valve” of capital-account opening, capable of relieving the pressure of large capital inflows and outflows.

In addition, the pressure from offshore markets was an important reason the United States loosened capital controls, while the depth of the onshore market provided a key condition for the United States to open its capital account. Only through the coordinated development of onshore and offshore markets can a solid and sustainable institutional foundation be provided for renminbi internationalization. In the onshore market, market depth can be enhanced by improving the liquidity of the secondary market for Chinese government bonds and enriching interest-rate risk-hedging tools; in the offshore market, efforts should focus on developing the Hong Kong offshore renminbi financial market, promoting offshore renminbi bond issuance, enriching financial products in the offshore renminbi market, and improving offshore renminbi market infrastructure.

► Seize opportunities to actively promote renminbi-denominated settlement and improve the “renminbi cycle” mechanism. The reason funds did not evolve into long-term large-scale capital flight after the United States opened its capital account lies chiefly in the “petrodollar” mechanism and in the asset appeal brought by U.S. price stability and strong economic growth in the 1980s, which together drove the formation of the “dollar cycle.”

For China, to form a “renminbi cycle,” it should first broaden the channels for exporting renminbi liquidity overseas. Obstfeld (2026) pointed out that the export of dollar liquidity does not necessarily occur through a current-account deficit; it can also occur through outward investment via the capital and financial accounts, or through credit creation in the offshore dollar market. In the 1960s, the United States was still running a current-account surplus, but the offshore dollar market was already able to derive dollar liquidity overseas through the banking system. Similarly, while maintaining a current-account surplus, China can also export renminbi liquidity through the “Belt and Road,” enterprises going overseas, outward investment, and other means.

In addition, it can also seize the opportunities brought by external shocks and global industrial transformation, actively promoting renminbi-denominated settlement and the reflux of renminbi funds. China is the world’s largest manufacturing center, an important importer of energy and commodities, and has strong industrial influence in fields such as new energy, strategic minerals, and green industries. In the future, it can explore raising renminbi settlement and pricing in key resource goods such as strategic minerals, the “new three” [New energy vehicles, lithium batteries, and photovoltaic products], and even computing power.

Finally, stable prices, deep financial markets, and stable growth are key to attracting funds to continuously flow back. Overseas entities being willing to accept renminbi is only the first step of renminbi internationalization; being willing to invest in and hold renminbi assets over the long term is the key to the renminbi moving from a settlement currency to an investment currency and a reserve currency.

The reason the United States attracted funds to flow back after opening in the 1980s still lay chiefly in the asset appeal brought together by the price stability following Volcker’s inflation control and the growth vitality unleashed by Reagan’s supply-side economics. In the long run, only by building an internal ecosystem with stable prices, more resilient growth, sufficient market depth, and attractive asset returns can funds be attracted to flow in sustainably over the long term amid the dynamic allocation of global capital.

What stands out most is that the article relies on “exclusive disclosures from sources” to suggest that China could launch anti-discrimination and supply chain security investigations into the EU’s practices. The use of unnamed “sources” typically signals an unofficial backchannel, suggesting this is not the journalist’s personal speculation but rather a message to Brussels that, if the “overcapacity instrument” actually moves forward, these are the tools China is prepared to use in response.

Equally telling is that, in the following paragraph, the article goes out of its way to single out several EU exports to China. French cosmetics come first, with the article emphasizing that France has been the top source of China’s cosmetics imports for three consecutive years, accounting for 29.6% of total cosmetics imports in 2025. Meat, wine, and luxury goods are mentioned next. In my view, these product categories all command a significant share of the Chinese market, and their sources are concentrated in core EU member states such as France, Italy, Germany, and Spain. They are also “national flagship industries” in the countries concerned, tied to substantial employment and local political interests.

If the EU’s logic is that “production capacity exceeding domestic demand” qualifies as overcapacity, then these flagship European exports to China fit the same definition just as well, and could naturally become targets for countermeasures. Taken together with China’s earlier anti-dumping investigation into French cognac, the signal the article is sending, in my reading, is that the cost of pushing this legislation forward will fall first on the EU member states’ own pillar industries, and it is time to hit the brakes within the European Commission.

Below is the full article:

Exclusive Disclosure: If the EU Insists on Pushing the “Overcapacity Tool,” China Will Retaliate

On the 29th, the European Commission is set to convene a meeting to explore adjustments to its economic and trade policy toward China.

Just days before the meeting began, several EU member states submitted a proposal seeking to push forward the “Resilience Tool”—the so-called “Overcapacity Tool.”

While the EU has not stated it explicitly, this new tool is strongly targeted at China by design.

As early as 2023, the European Commission began hyping up the “China overcapacity theory.” Since the beginning of this year, whether it is the “Overcapacity Tool,” the draft amendment to the Cybersecurity Act, or the recently released Industrial Accelerator Act, all signify that protectionism is being directly entrenched by the EU into its economic and trade policy framework toward China.

The EU’s economic and trade policy toward China is moving further and further down a radical path.

So, where does this radical turn come from? And what impact will it have on the EU itself?

Some EU officials believe that traditional anti-dumping and anti-subsidy investigations take too long, and that it is more efficient to directly preset quotas—imposing tariffs as soon as the quota is exceeded, which would be faster and more efficient.

This means that once a country’s products exceed the limit, the EU can systematically attack all related industries of that specific country under the pretext of “overcapacity.” Any industry with high market share and competitiveness in the EU market could become a target.

The conceptual logic applied here is full of holes. To understand why the EU is accelerating its turn toward protectionism at this moment, one must extend the timeline.

It must be recognized that this is not a momentary impulse on the EU’s part. Over the past two decades, Europe’s industrial competitiveness has continued to decline, and now this “chronic disease” has deteriorated to the point where extreme measures must be taken.

From a macro data perspective, European industrial production had just strongly recovered from the pandemic in 2021, with industrial production growth reaching 8.5%; in 2022, the year-on-year growth rate plummeted to 0.3%, almost stagnant; in 2023, it fell 1.4% year-on-year; in 2024, it fell 2% year-on-year.

Why did European industrial production growth suddenly hit the brakes in 2022? The answer lies in the energy bill.

Europe’s benchmark natural gas price soared from about €20/MWh in early 2021 to nearly €340/MWh in the summer of 2022.

Over the following three years, although the EU worked hard to diversify its energy imports, energy costs failed to fall back to low levels. In 2023, EU industrial electricity prices were about 158% higher than those in the United States, and natural gas prices were 345% higher; this gap had still not significantly narrowed by 2026. In other words, this is not a one-time price fluctuation, but a structural wound.

When energy costs remained high, the defenses of European industry began to collapse.

The chemical industry is the EU’s fourth-largest manufacturing sector, long known as “the industry of industries,” with its products penetrating 95% of manufacturing categories. From automobiles to medicine, from defense to daily necessities—once the chemical industry falls, the entire industrial chain will be affected.

And the data for the chemical industry presents precisely a collapse-style decline.

Data from the European Chemical Industry Council shows: chemical production capacity announced for shutdown in 2022 was 2.9 million tons; in 2023 it surged to 8.7 million tons; in 2024 it remained at a high level of 8 million tons; in 2025 it further skyrocketed to 17.2 million tons.

Over four years, cumulative lost capacity totaled 37 million tons, directly leading to 20,000 job losses, with nearly 90,000 indirect jobs at risk—the industry is on the “brink of collapse.”

At the same time, the decline of these industries is spreading to other sectors: the construction industry continues to decline, and the aluminum industry has been almost uprooted—the EU’s primary aluminum production is only 950,000 tons, while annual consumption is as high as 13.5 million tons, with a structural gap of 93%.

Together, these industries paint a bleak picture: a complete energy-intensive industrial chain is quietly disintegrating.

It can be seen that the root cause of the EU’s industrial predicament is structural, and its manifestation at the trade level is the widening trade deficit.

As the China Finance 40 Forum’s research concludes: the widening deficit is not the spillover pressure of so-called Chinese “overcapacity,” but rather “the combined result of the upgrading of Chinese manufacturing and Europe’s energy constraints.”

In other words, it is not that Chinese industry has “attacked” Europe, but rather that Europe itself has fallen into difficulty.

However, we have seen that since 2023, the European Commission has been trying at the level of economic and trade policymaking to deflect contradictions in order to avoid the truly difficult structural problems.

According to a report by the European Council on Foreign Relations, the EU is aware that whether rebuilding supply chains or cultivating domestic industries, it will take years to see results, while real-world shocks are often immediate.

Therefore, the report argues that only by adopting tough confrontational measures can the EU gain the initiative in economic and trade competition, and uses this as an excuse to accelerate the turn toward trade protectionism.

Last December, the document “Strengthening EU Economic Security” explicitly proposed shifting “from a reactive posture to a more proactive and systematic use of various policy tools.” Even in some cases, “the EU, its member states, and industry will increasingly need to be prepared to bear corresponding economic costs in order to reduce external dependence and enhance overall security levels.”

This was followed by the successive appearance of documents such as this year’s Industrial Accelerator Act.

Taking the Industrial Accelerator Act as an example, this act claims to be industrial support, but in reality wraps layer upon layer of discriminatory provisions, setting up multiple barriers against foreign enterprises in the four major industries of batteries, electric vehicles, photovoltaics, and critical raw materials.

Its main proponent is Stéphane Séjourné, Executive Vice-President of the European Commission in charge of industrial strategy, who is also a representative figure of EU protectionism. And people like him are not few among current EU economic and trade policymakers. Behind these policymakers, an important force is surging.

Related reports show that within one year of the launch of the Clean Industry Deal, more than 750 lobbying sessions were held around it—an average of more than three per working day.

Among them, the most active are not new energy enterprises, but traditional heavy industry giants in steel, energy, automobiles, cement, and so on. Séjourné and his team are among the most frequently lobbied targets, having held 131 meetings with 192 lobbying groups.

These industries, without exception, are the sectors that find it most difficult to pivot amid the EU’s structural economic predicament.

However, for an EU economy in urgent need of structural transformation, engaging in protectionism according to the demands of these traditional industries runs completely counter to the direction of energy structure and industrial structure transformation—it is essentially drinking poison to quench thirst.

This is a vicious cycle—the more difficulties internally, the more protection; the more protection, the less impetus for transformation, and the deeper the entanglement in the predicament.

Based on the above analysis, several conclusions can be drawn.

First, the EU’s turn toward trade protectionism is essentially the result of the long-term decline of European industry colluding with the lobbying of vested interest groups. The structural disadvantages in energy costs, the continuous shrinking of manufacturing... these chronic ailments accumulated over two decades have erupted at an accelerated pace after 2022.

Faced with the predicament, the EU has not chosen to “scrape the poison from the bone” but instead uses layer upon layer of barriers to avoid reform. From the draft amendment to the Cybersecurity Act to the Industrial Accelerator Act, from the “Overcapacity Tool” to the Foreign Subsidies Regulation—standing behind all of these are both politicians who accumulate political capital through “tough narratives” and traditional industrial giants worried about being eliminated by the green transition. Preserving market share and defending vested interests is the real script of this drama.

Second, China’s countermeasures are not verbal warnings.

According to exclusive disclosures from sources, China can launch anti-discrimination investigations and supply chain security investigations into the relevant EU practices. The Ministry of Commerce has made it clear that once China’s national interests and the rights and interests of its enterprises are harmed, China will resolutely take countermeasures.

If the EU insists on pushing forward the so-called “Overcapacity Tool,” China will immediately take action and adopt comprehensive countermeasures. China is neither unfamiliar with nor afraid of trade frictions, and will see it through to the end.

Third, the logic of the EU’s accusations does not hold up in itself.

For three consecutive years, France has been China’s largest source of cosmetics imports, accounting for 29.6% of China’s total cosmetics imports in 2025. EU exports to China—meat, alcohol, luxury goods, cosmetics, and other products—all occupy significant shares of the Chinese market.

If the EU accuses Chinese products of “overcapacity” using the absurd logic of “production capacity exceeding domestic demand,” then whether these European products sold in China face the same situation is also worth considering.

The session focused on the trajectory of China-U.S. relations after the Xi-Trump meeting, the prospects for managing strategic competition, and the implications of “a constructive bilateral relationship of strategic stability”.

The event was moderated by Mabel Lu Miao, Co-founder and Secretary-General of CCG, and drew diplomats, scholars, journalists, and representatives from international organisations, think tanks, and the business community.

During the Q&A session, Vebjørn Dysvik, Norwegian Ambassador to China; Marcelo Suárez Salvia, Argentine Ambassador to China; Colum Murphy, Senior China Economy and Government Reporter at Bloomberg News; Zhuorui Fu, Associate Economic Officer, UN ESCWA; Shi Yinhong, retired Professor at the School of International Studies, Renmin University of China; and Zhao Shengchuan, Professor at Josai International University, raised questions and exchanged views with Shambaugh and Wang.

This transcript is based on the event recording and has not been reviewed by any speaker.

Mabel Lu Miao, Co-founder & Secretary-General, CCG

Your Excellencies, ladies and gentlemen, friends and colleagues, good afternoon. A very warm welcome to the Center for China and Globalization, and to this special edition of our CCG Global Dialogues series.

My name is Mabel Lu Miao, Co-founder and Secretary-General of the Center for China and Globalization. First of all, I thank you all for your presence as we gather to explore one of the most consequential relationships of our time: the future of U.S.-China relations.

We meet at a pivotal moment. The dynamics between the United States and China are shaping not only the bilateral landscape, but also the very architecture of global stability, trade, and innovation. Following the recent high-level summit between President Xi and President Trump, we find ourselves at what many call a critical crossroads. The consensus to build a constructive relationship of strategic stability offers a new framework, a diplomatic anchor in turbulent seas. Yet the fundamental forces of strategic rivalry and deep economic interdependence continue to coexist, defining a relationship that is as complex as it is crucial.

The question before us today is not merely academic; it is urgent and profoundly practical in an era of persistent contestation: How can these two great powers stabilise their interaction? Where are the realistic avenues to expand cooperation on global challenges, from climate change to public health? And perhaps most pressingly, how can both sides effectively manage risks and prevent miscalculation amid inevitable competition?

These are the questions that today’s dialogue seeks to address. To guide us, we are honoured to welcome a scholar whose work has illuminated the U.S.-China relationship for decades. It is my great pleasure to introduce our distinguished keynote speaker, Professor David Shambaugh.

David Shambaugh is Gaston Sigur Professor of Asian Studies, Political Science, and International Affairs at George Washington University’s Elliott School of International Affairs. He is also a distinguished visiting fellow of the Hoover Institution. Professor Shambaugh is an internationally renowned scholar on contemporary China, Asian affairs, and, of course, U.S.-China relations, with a career spanning academia, policy analysis, and extensive field research. He brings a rare blend of theoretical depth and practical insight. His perspectives have informed policymakers, business leaders, and scholars across the globe.

We are privileged to have him with us today to set the stage for our discussion. Following his keynote, Professor Shambaugh will be joined for an in-depth dialogue with Henry Huiyao Wang, Founder and President of the Center for China and Globalization, former Counsellor of the China State Council, and a professor at China Foreign Affairs University. Henry Huiyao Wang has dedicated his career to fostering international exchange and think tank diplomacy, engaging in profound dialogues with world leaders and thinkers. His unique global perspective will help bridge theory and practice, ensuring our discussion is both grounded and forward-looking.

Our agenda today is designed for substance and interaction. After the dialogue segment, we will open the floor for your questions, because the insights from this distinguished audience, comprising policymakers, senior scholars, business leaders, and journalists, are an essential part of this global conversation.

Before we begin, I would like to extend my sincere welcome to our distinguished guests, including His Excellency Marcelo Suárez Salvia, Ambassador of Argentina to China. Welcome, Your Excellency. His Excellency Vebjørn Dysvik, Ambassador of Norway to China. Welcome. Also, Ms. Amakobe Sande, UNICEF Representative to China and Professor Shi Yinhong from Renmin University. We also warmly welcome scholars, diplomatic officials from various embassies, business delegates, and media friends. Thank you all for your presence.

Let us begin this important conversation now. Please join me in giving a warm welcome to our keynote speaker, Professor David Shambaugh. Welcome.

David Shambaugh, Gaston Sigur Professor of Asian Studies, Political Science, and International Affairs, George Washington University; Distinguished Visiting Fellow, Hoover Institution

Henry and Mabel, colleagues and friends, 各位观众, distinguished members of the diplomatic corps here in Beijing, I am very honoured, actually, that all of you came out on a rainy day in Beijing. But thank you for making the effort. I know getting around Beijing is not the easiest thing with the traffic, although I think traffic has improved slightly from the last time I was here. The air has also improved considerably.

Thank you for coming. It is truly a great pleasure for me, and that is an understatement, to return to Beijing, a city I first visited in 1979 when I was a student, and have lived in myself for six years altogether. This is indeed another home of mine, I might say. But for reasons I am not going to go into, this is only my second visit in the past 11 years. Therefore, I am sincerely grateful to Henry, to Mabel, and to CCG for the invitation to return and for making this visit possible.

I truly hope that my long absence is now firmly in the past, and that I can look forward, hopefully, to many more visits in the future, resuming my longstanding relationships with the many institutions and friends here in Beijing and in China, and making new friends, particularly with the younger generation.

I have been coming to China every year from 1979 to 2015. That is a long time. I have studied at three different Chinese universities as a student. I have been a visiting scholar at six different institutes of the Chinese Academy of Social Sciences, and other universities and think tanks. We have a long and good past. It is wonderful to be back, and I am really very grateful.

I would also like to begin by paying tribute to CCG, in all sincerity, for the extremely important role that you have played for the past 18 years since your establishment in providing multifaceted forums for dialogue between China and the world, both inside China and, of course, around the world. I think Henry and Mabel live on airplanes. I do not know how you travel as much as you do. I receive your CCG activity emails in my inbox, as I am sure many of you do. One day they are in Latin America, the next day they are in Europe, the next day they are in Southeast Asia, then they are back in Beijing, and then they are off to some other place. I do not know how you do it.

But in all sincerity, I want to congratulate you on the forum that you established 18 years ago, and on how it has grown into what it is today. In fact, just this morning over breakfast, I was reading your annual report, which, by the way, is here on the table. If any of you need copies, I commend you to pick one up on your way out. It is very impressive in its range of activities. As they say in Chinese, 加油 keep it up. You really deserve congratulations for what you have created, established, and developed. People-to-people exchanges among thought leaders and other elements of societies are really more critical than ever before in today’s very turbulent world.

I am going to speak for about 20 minutes, I am told, about a very big subject: U.S.-China relations, the state and future of the relationship following the very recent summit meeting here in Beijing between our two presidents. So, let me say a little bit about how I view the results of the summit and the possible evolution of the relationship going forward. Then Henry and I are going to have a constructive, strategic, and stable dialogue.

中美建设性战略稳定关系 is the new 提法 (tifa). We have a new tifa. For those of you who speak Chinese, tifa means narrative. There is a new narrative to define the U.S.-China relationship, an official narrative, which is very important because policies derive from narratives. Narratives are like a grand strategy, 大战略 in Chinese. We have a new tifa coming out of this last summit: to build a constructive, strategic, and stable U.S.-China relationship. It is the first time those terms have been used in these narratives.

First, before I talk a little bit about the narrative and unpack it, let me give you my bottom-line takeaways and evaluations. Let me talk a little bit about the summit, because I am sure you have all been following it and digesting the tsunami of information that came forth during and susequent to the summit. It actually takes a number of days, oftentimes weeks, before you really get to understand what happened and what came out of a summit. We are beginning, day by day, to learn more. It occurred just a week ago, so I do not think we have heard everything yet. I have no inside 消息, no inside knowledge.

There were relatively few concrete and specific deliverables, as Americans call them, that came out of the summit. There was no joint communique, no joint agreements, and very little tangible that one could identify in the immediate aftermath of the summit. Day by day, we are learning a little more. But that is not the most important thing.

In Washington, where I live, there has been some criticism of the summit because it did not achieve so-called concrete deliverables. Americans are very concrete. They like specifics; they do not like abstract phrases. They want concrete deliverables. This summit did not produce concrete deliverables, so there has been a bit of criticism and frustration in D.C. I have to be honest with you.

But for me, the greatest value and the most important outcome was that the summit occurred at all. It is absolutely crucial that our two governments and top leaders meet, and do so with some frequency. This is just the first of four meetings this year. They have three more to go. There will be a return state visit by President Xi, General Secretary Xi, to the United States in September, a meeting in Shenzhen, and a meeting in Miami. So this is really important.

This particular summit succeeded in putting a floor, I think, under a very fluctuating, rocky, and troubled relationship. It has not been in free fall; I would not say that. A few years ago, in the second year of the Biden administration, it was in free fall. So the Biden administration tried to put a floor under it and arrest the deterioration. This time we needed a floor to be relaid, you might say, and that is the major accomplishment to my mind. A floor was relaid, as well as establishing some guardrails and boundaries, both of which will serve to stabilise relations and, I hope, begin to re-establish a more lasting framework to manage the competition.

“Manage the competition” is a particular American phrase. The Chinese side does not like that phrase and does not use that phrase specifically. But the summit allows us, I think, to have a new framework to restrain and limit points of friction and keep the elements of contention, including the all-important issue of Taiwan, within manageable boundaries, hopefully.

So I view this summit as having been quite successful. Let me be very clear about that. I am not disappointed by it, because I think the signals that were sent, both the optics of the summit as well as what we have learned about the substance. So I think it might hopefully be a turning point in the relationship. I have been studying U.S.-China relations long enough to know that there is never any real turning point. We never get around the corner and stay there in a stable place. But this is a new attempt, and it is the best attempt in about 10 years, I think. Let us hope it endures.

The United States and China are not going to stop competing and having points of friction, many points of friction. But I think it is important to understand what competition means. Let me digress a little onto that subject.

What I have found most instructive recently in terms of understanding competition is the May 17 press conference and media briefing by State Councillor and Foreign Minister Wang Yi. He gave a very extensive readout on the summit. You should all read it. It is four or five pages long. On the subject of competition specifically, let me quote briefly what State Councillor and Foreign Minister Wang had to say, because I think it is very useful and instructive.

For the last six years, the Chinese side has refused to even engage with the Americans over the term competition. They thought it was some sort of negative, hostile conflict. You only compete with adversaries. But now the Chinese side seems to have come around to not viewing competition as an adversarial phenomenon, which is not.

Foreign Minister Wang said, and I quote, competition should be healthy and stable, “where competition is kept within proper limits and is not turned into a zero-sum game.” “Major-country competition is nothing new,” he said. “When competition does happen, it must be a healthy one where we learn from each other, pursue excellence together, and compete fairly in compliance with rules.” “The purpose of competition,” Foreign Minister Wang said, “should be to outdo oneself, so that both sides become better.” That is a positive-sum definition of competition.

Then he concluded that “it should be constant stability where differences are manageable, and the relationship should not be like a roller coaster”. Notice the word “manage”, 管理. We Americans have been working on the Chinese side with this concept of managed competition for six years. It seems to have finally sunk in a little bit.

I completely agree with Foreign Minister Wang’s characterisation, and I commend him on it. The entire press conference is worth reading. I just read a little bit. Competition is an entirely natural 自然的 part of life in many domains: in commerce and economics, in sports, in the realm of ideas and even ideologies, and in geostrategic affairs. So there is nothing abnormal about competition. Quite to the contrary, as Foreign Minister Wang said, it can spur one to improve themselves and to compete with others, and that in itself is a positive-sum phenomenon. Now, to effectively manage competition necessitates direct exchanges at many levels and candid dialogues. So I think that was a very useful contribution by Foreign Minister Wang.

He also said that President Xi raised with President Trump the topic of the Thucydides Trap. That is interesting. It is not clear to me that President Trump has read Graham Allison’s book and even understands the notion of a Thucydides Trap, but obviously President Xi has, and has had personal discussions with Professor Allison.

This is important, I think, because his book, which had a very bad title, Destined for War—believe me, I know about bad titles. Publishers sometimes give articles and books very provocative, inappropriate titles with which the author does not share. Believe me, that has happened to me personally. Personally, that is one of the reasons I have not been here for 10 years. The Wall Street Journal gave a very inappropriate title to an article I wrote 10 years ago.

Professor Allison’s book looks at, I think, 16 historical cases of rising powers and established powers, going back to Athens and Sparta. He found historically, 14 of the 16 cases resulted in conflict, usually because the established power was feeling threatened. Sometimes the rising power attacked the established power early because they thought they had an opportunity. Sometimes the established power attacked the rising power to keep it from parity and from becoming a peer competitor. Anyway, in 14 of 16 historical cases, the historical record is pretty clear and very dangerous.

I think it was very appropriate that President Xi raised this with President Trump. But the point here is that nothing is structurally predetermined in international affairs. In political science, in what I do, we call this the agent-structure paradigm. Leaders possess agency, and the exercise of human will can ameliorate and overcome structural dynamics that produce tensions in a relationship. So nothing is predetermined, and I hope President Trump took that away from his discussions.

The last thing that Wang Yi said I found very useful was that he reported, among the other topics discussed, people-to-people exchanges should be increased. Indeed, they do need to be increased. These have atrophied badly between the United States and China during and since Covid, and we need to rebuild people-to-people exchanges on a reciprocal basis.

Cutting off dialogues, or worse yet, banning people from the other country, is not helpful. In fact, it is harmful. We need to be more tolerant and open to various perspectives. Again, this is what CCG, Henry, and Mabel do so well.

People-to-people dialogue needs to be rebuilt. And I hope that includes not just tourists and students. Students are very important. President Xi reiterated his invitation to 50,000 American students to come to China over a five-year period. That is a very nice aspirational number, but it is going to be very hard to achieve, especially given that today, as we sit here, there are only 1,200 American students in this entire country. Ten years ago, there were 20,000 American students in this country. There are, by contrast, 270,000 Chinese students in the United States this academic year, including in my classrooms. We have got to work on that.

But I would also like to suggest that we need to work on improving scholarly exchange, not just student exchange—scholars who already have their PhDs and are already researching different topics. There are so many interesting things going on here in China that the international China studies community, not just the American China studies community, is very anxious and curious about. We all want to come here. We want to do research in archives, libraries, factories, laboratories, and travel in the countryside. I hope that when the two governments get around to implementing the people-to-people component, that they really pay attention to the scholarly component.

I am relatively positive about the recent summit. It was very businesslike and obviously very formal. It was a state visit, and the Chinese side puts on such ceremony like hardly any other country in the world so well. It was a very respectful and polite summit, and there was even personal warmth between the two leaders. It was very obvious and good to watch.

We have this new framework to constructively stabilise and develop a strategically stable relationship. Foreign Minister Wang—I am not going to go into it—also identified eight specific areas. So, again, I commend you to look at his May 17 press conference. The United States and China, according to him, are going to improve exchanges on foreign policy, military-to-military relations (hugely important), the economy and trade, public health, agriculture, tourism, people-to-people ties, and law enforcement. Those are the eight priorities. I think those are the right priorities.

Let us move back to the new tifa, the new narrative, and unpack that a little bit. I will be very interested to hear Henry’s observations on this new phraseology. Both governments have officially committed themselves to this as the guiding framework for the relationship.

I personally am not the biggest fan of slogans and narratives in international relations to define relationships, but I think they can be very useful at certain times. I see the utility in them because they serve as a kind of compass to guide a relationship, even a road map, you might say, to keep it more or less going in the same general direction, what the Chinese call the 主要方向, so you do not get off the road and have an accident or run out of petrol. But if you have a Chinese EV, you will not run out of petrol.

I think this new phrase is better than previous phrases. I was not a particular fan of the 新型大国关系 [new type of major-country relations] that was announced between President Xi and President Obama many years ago. But we have a new one now, so let us talk about this.

Given the extreme complexity and the various systemic elements of the U.S.-China competition, stability is indeed a very worthy and appropriate goal for the two sides to work towards. Nonetheless, as in economics, equilibriums are hard to sustain. True stability is very difficult to maintain.

And I recall, in this context, my good friends and colleagues Professor Wang Jisi of Peking University and Kenneth Lieberthal of the University of Michigan, who was one of my mentors, and Jisi is one of my mentors and friends. They wrote an article about 10 years ago for the Brookings Institution in which they used the term “dynamic stability”. [Transriber’s note: The speaker appears to be referring to Lieberthal and Wang Jisi’s 2012 Brookings monograph, although the term “dynamic stability” does not appear in that work.]

In other words, you should not think of stability as something static and fixed. That is not the way to think of stability. It is unrealistic, not something where you achieve stasis, right? Life, all of us in our relationships, personal and between countries, evolves. Relationships are affected by all kinds of factors all the time. The question is how to create a framework that permits evolution, you might say, while working to achieve basic equilibrium. I think that is the first thing I would say about the 稳定 stability factor. Stability is not stasis. You have to think of it as a more fluid concept.

What about the term constructive, 建设性? That is an interesting term. When it first came out, it made me think about the history of the People’s Republic of China. There is a certain period of the 1950s known as the 建设时期, the build-the-country period. But that is not what they have in mind here.

I think “constructive”, first of all, is different from “cooperative”. Do not equate constructive with cooperation. They are two different concepts. Constructiveness indicates intention, and the intention to act and speak constructively indicates a kind of positive intention towards another party. Even criticism—in English, we have a common saying, “constructive criticism”. You all have heard that term. That means you can criticise someone if it is intended constructively for them to sort of self-reflect a little bit on whatever, and to maybe alter their thinking. That is an important point. Not all people or societies take criticism constructively. Some are reflexive; they just cannot take any criticism. I am not identifying anybody in this context. The point is to have an open mind and take criticism constructively, because it is positively intended.

I have spent my entire professional career studying the role of mutual perceptions in U.S.-China relations, beginning when I was a graduate student here at Peking University from 1983 to 1985. That is a long time ago, folks. I researched and wrote my PhD dissertation here. I was in fact the first foreign, not just American, student permitted by the Ministry of Education to study in an international relations department in China in 1983. They opened the door, and I was the first one to walk through that door. Many more have walked through the door subsequently. It was a wonderful experience, two years in the 国政系. I also played for two years on the Beida [Peking University] basketball team, and we had a lot of success. We won two city championships. I played in the Chinese national basketball tournament for two years. I had great experiences. I have wonderful memories of Beida, and previously Nankai University and Fudan University.

The reason I am telling you this, sorry about the tangent, is because I was researching a dissertation on China’s America specialists and their perceptions of America, 中国的美国问题专家看美国知觉. I am an American China specialist, so I wanted to look at my counterparts: What about China’s America specialists? What are they thinking about the United States? How do they understand the United States? That was my doctoral dissertation almost 40 years ago, and it turned into a book titled Beautiful Imperialist, 美帝 (mei di)—mei for America, di for imperialism—because Chinese views of America at that time were ambivalent. On the one hand, America was to be emulated, studied, and learned from. On the other hand, America was imperialist and should be rejected and criticised. So there is this ambivalent tension Chinese views of America for a long time. It goes back even earlier than the 1980s.

The reason I raise this is that if I have learned nothing more in my decades of study of perceptions in the U.S.-China relationship, it is that it is actually THE most important factor, much more important than the trade balance or other functional dimensions of the relationship. We need to understand each other better. And language means different things to different people. One should never assume that others understand your language and see things the way you do.

This is precisely why sustained dialogue between our two societies, and all societies, is crucially important. All of you representing other countries here in China, I know, have the same view. Here again, I wish to pay real tribute to Henry, Mabel, and CCG for what you provide in China’s dialogue with the world. There is no institution like this in China other than CCG. Very impressive. Keep up the good work.

Last point, and then I should stop and we should have our dialogue. What about the term strategic? Constructive, strategic, stability: those are the three operative adjectives and nouns in this new tifa. Strategic is interesting because there are differences of definition that need to be further explored through dialogue. For Americans, I can tell you, the term strategic usually, if not always, refers to the security and military domain. Full stop. Security, military equals strategic for Americans. That is how we think. Not for others.

The Chinese—and I have been reading these articles in the last week after the summit—many Chinese scholars are writing very interestingly that, oh no, 战略 strategic is a kind of comprehensive, elastic term that encompasses all dimensions of a relationship, and therefore it’s strategic; all dimensions equal strategic. Well, that is a big difference in the way Americans and Chinese think about that term. I think we need to have some dialogues to unpack our thinking about “strategic” and about “competition”.

We are making progress on competition. As I said, I think Foreign Minister Wang’s statements this week, and even at the Munich Security Conference, show a lot of evolution in Chinese thinking. I am not going to say progress, but it is just evolution. So it is good.

So, Henry, keep it up. We need to have more of these kinds of dialogues. I have gone on long enough, but those are my general impressions of the summit and the state of the relationship. Henry, you and I can now sit down and have a constructive, strategic, and stable dialogue. Thank you very much.

Henry Huiyao Wang, Founder & President, CCG

As we all know, as Mabel just introduced, David is a very famous and renowned China expert in the U.S. As you said, you studied in China for so many years and did all those studies. Of course, you went back to the U.S., and you were also the chief editor of The China Quarterly for almost a decade. That was really a very prestigious China research journal, a top journal in the world, while you were in the UK for 10 years. You have also published many books.

I know you came to Beijing two years ago with a Central Party School Track II dialogue, where I was also participating. I am glad to see you come back again, and, as you put it, to promote better dialogue and understanding.

I am very glad to have you join us for CCG Global Dialogue. We have had almost 100 dialogues, including with a few dozen top American scholars, such as Joseph Nye, Graham Allison, Tom Friedman, Jeffrey Sachs, and many more. Now we are adding you to this long list of well-known strategic global dialogues.

Your timing is very significant. You came just one week after the summit. My first question and comment to you is that this visit, I personally feel, is very significant. It is almost a decade—nine years—since Trump’s last visit. This is his second visit. You can see that President Trump used much more positive terms with the Chinese side. He basically praised what China has done and expressed many good gestures towards China. I think that has been appreciated.

At the same time, as a Chinese person, I feel China is now getting a lot of international attention. Since the beginning of this year, we have had almost all G7 leaders come, except Japan. At the end of last year we had President Macron, and at the beginning of this year, Mark Carney, then Starmer, then Merz, and almost all of them. Then after that, Putin. You see those big powers all coming, and that really shows that China has become a focal point, an internationally attractive assembly place where world leaders come to compare notes and build consensus.

How do you think China has reached this kind of influence and impact in the world, with almost all G7 leaders and other global leaders coming? Also, we are in a much more turbulent world now. We still have a war going on in Ukraine. We have a terrible war going on in Iran. The whole world has these choking points. What do you think about this? This is the first U.S. state visit in nine years. During the four years of Biden, as you rightly said, we never saw a U.S. official visit China, which is unthinkable for the two largest economies in the world. There was no dialogue or exchange.

Now this high-level exchange has resumed. We are going to see another President Xi visit in September, another APEC summit in Shenzhen, and the G20 summit in Miami. Could 2026 be a turning point, as you just mentioned, between China and the U.S., and perhaps another turning point for China, the West, and the world as well? Maybe you can give some of your thinking on how special 2026 can be, or what China has done right.

David Shambaugh

Thank you, Henry. There were a number of elements in there, and I will see if I can unpack them. But your main point about foreign heads of state and dignitaries increasingly coming to China is a very well-taken point. That is not necessarily a new development. China’s diplomacy, in quantitative terms, is really unparalleled, both here in Beijing and in travel by President Xi. President Xi is not travelling as much in recent years as he used to, but I was reading the other day that he has made 71 trips abroad to 112 countries during the time he has been in office.

Anyway, I think Beijing and China are becoming more of a destination point for various heads of state. Part of that is the result of the Trump factor. We know what has been happening within Europe in particular: very unfortunate transatlantic policies and verbiage. So, several European heads of state have beaten a path here to Beijing. Canadian Prime Minister Mark Carney came here and delivered a very significant address in Davos about middle powers.

So China is reaping some of the benefit of the alienation that American allies and others are feeling about the United States under President Trump. There is a kind of, you might call it, a bump or an effect—the Trump bump. How is that for a phrase?

Nonetheless, even without Trump, other countries are hedging. I just published an article in Foreign Affairs last week about how the whole world is hedging against both America and China. Countries around the world do not trust either power. They do not trust China, they do not trust America, and they certainly do not trust Russia. Everybody is hedging, and part of hedging is in fact engagement.

You just had President Putin here last week. I think China is playing its diplomatic role very effectively at the moment, and will continue to. That is important. For the rest of 2026, there are going to be these multilateral meetings. I will give President Xi and China real credit here in the realm of global governance. China has really stepped up its game in the global governance domain under President Xi.

I teach about this in my Chinese foreign policy classes. Up until President Xi became the leader of China, China was very ambivalent about multilateralism. Verbally, they were supportive, but in practice not. There has been a much greater convergence of China’s behaviour and participation with the verbiage and language, which has always been supportive of multipolarism.

Ladies and gentlemen, multipolarism and multilateralism are not the same thing. International relations specialists will tell you that polarism has to do with poles, with powers. Powers attract other countries to them like magnets. That is what a power is. Multilateralism is institutional: the United Nations and all these other IGOs in the world.

China is now playing a much more central role—the new Global Governance Initiative that China just published three or four months ago. I was along with Xi Jinping’s Global Civilisation Initiative, Global Security Initiative, and Global Development Initiative. Now we have a Global Governance Initiative. These are all indications of China’s increased activism—not just centrality, but activism in world affairs. This is very positive. Very positive. So I think we can anticipate that, going into the end of 2026 and into next year, that that trend will continue.

Henry Huiyao Wang

Thank you. Yes, David, absolutely. You see China becoming more active, particularly in turbulent times, making many proposals and initiatives, particularly for global governance, where we are actually embracing a multipolar world. Global governance is really at the heart of the issue. This centrality that you mentioned is gaining some momentum now.

Again, you mentioned during your speech that I was quite delighted to hear that the U.S. and China, as President Xi put it, have reached constructive stability and strategic stability. I know the U.S. added “on a reciprocal basis”. But I remember that in 2017, when the first national security strategy report came out under the first Trump administration, strategic rivalry was really the buzzword. Strategic rivalry was even before Russia. So this turnaround, from Trump 1.0 strategic rivalry to Trump 2.0 strategic stability, is really a big shift over the last decade.

The first Trump term started with a trade war, a tariff war, and sanctions. Even today, there are about 2,000 Chinese companies on the American entity and sanctions lists. But that profile does not really work that effectively. As a matter of fact, I had a dialogue with the late Joseph Nye about a year or two ago. He told me that there is a cycle in U.S.-China relations roughly every 20 years. China and the U.S. were at war in the Korean War, that was the lowest point. Twenty years later, the Nixon moment came, and then there was a revival, followed by another downturn.

Are we now seeing the bottom? Are we starting to realise that we cannot really decouple? One of the things I noticed when I went to the United States last month for a world economy summit with 500 CEOs was that I never heard anybody talk about decoupling with China. Even senior U.S. officials were saying that the U.S. does not seek decoupling now.

In that sense, are we achieving this kind of realisation because we have achieved mutually assured disruption? From MAD, mutually assured deterrence, and mutually assured disruption, to what you just said: mutually assured construction. By that kind of realisation, after nine years, we figure out that as wars are going on and the world is in a different chaotic place, the U.S. and China have to work together. We have to, as President Trump put it properly, recognise a G2 or near-peer status.

What we get from strategic stability, or what you have called for many years managed competition, is that we have to recognise reality. We cannot go on fighting each other or sanctioning each other. We have to find a way to coexist peacefully. Even though strategic stability may not be that concrete, as you said, in the Chinese narrative, if you have a guiding narrative, then everything else follows through and can be much easier. That is how Chinese people approach things: it is a long-term objective.

What do you think about this paradigm shift, from the first term’s strategic rivalry to the second term’s strategic stability? I hope we are going to bottom out and maybe keep managed competition, as you mentioned. Minister Wang mentioned Olympic-style competition. Maybe we could seek a more stable world as time goes on. What do you think about this paradigm shift, or this move from strategic rivalry, which the Trump administration no longer emphasises? I have not heard any official talk about strategic rivalry. On the contrary, I still hear Europeans talk about systemic rivalry, which probably was inspired by the first Trump term. But the U.S. is not mentioning strategic rivalry; now it talks about strategic stability. It adds some modifiers at the end, but it is still strategic stability. What is your take on that?

David Shambaugh

Again, these phrases and narratives are important for us to consider, but we should not put too much stock in them. Just because you declare something does not make something that way in reality. We can declare that the U.S.-China relationship is a strategically stable relationship, but in fact, that is aspirational, I would argue.

Henry Huiyao Wang

Could this also be a recognition of reality?

David Shambaugh

The reality is comprehensive competition, comprehensive across all domains and every region of the world.

Henry Huiyao Wang

But it has to be managed.

David Shambaugh

But it has to be managed, exactly. I was the first one to use the term “managed competition” about 12 years ago, Henry. That is interesting; you knew that. I also coined the term “competitive coexistence” six or seven years ago. That is what I think we should try and achieve in the U.S.-China relationship.

The U.S. and China are going to be competing across every domain and every region of the world as far as the eye can see. We need to recognise that reality. We should not deny it, but you have to learn how to manage it. There are different mechanisms for management, beginning with dialogue, but there are other institutional mechanisms for managing strategic rivalry, in the military domain, for example.

Let us be clear about why the narratives are important, because they set the general direction, the 主要方向, and they produce guardrails and a road map. That is all very positive. We did not have guardrails and a road map before last week. Throughout the Biden administration, they tried a little bit, but the Biden administration did not succeed very well. Now at least we have a framework, a floor under the relationship, and we have guardrails around it to a much better extent.

But look, these are the two big powers in the world. If you take International Relations 101, you start with Athens and Sparta, and you move on through late 19th-century Europe, into the 20th century in the Cold War, and into the 21st century. These two powers are competitive. They are intrinsically competitive in every domain. But that goes back to what I was saying earlier, that competition is not a negative concept. It is not a zero-sum phenomenon. It can be a positive-sum, constructive element in a relationship. So I guess that is how I would respond.

Henry Huiyao Wang

Great. You mentioned the Thucydides Trap. Actually, President Xi talked to President Trump during the meeting, saying that this trap does not have to happen and we do not have to follow that old path. I remember we had a dialogue with Graham Allison a few times. I even wrote a book on how to escape the Thucydides Trap, summarising our dialogues. Basically, I think the title of his first book is very frightening: Destined for War. We want to avoid that.

But peace in the whole world is in danger now. You see what happened in the Russia-Ukraine war, and what happened in Iran with the U.S. and Israel. Larger peace and the stability of the world are in jeopardy, while China is the largest trading partner with every one of them. That is why I understand why President Trump and Putin came to China, and also why the Pakistani prime minister is coming, and why the Iranian foreign minister was here not long ago. They all want to know where China stands.

If China says, for example, that the Strait of Hormuz has to be open, that war should not continue, and that dialogue must be sought, that is an enormous message China can give. I think Trump has gained some consensus from China on that as well.

On the other hand, Trump also said that the U.S. does not want to see Taiwan become independent and does not want to send soldiers 9,500 miles away. Do you think that, since the U.S. president wants to be a peace president, as he said in his inaugural speech, and now he is involved in wars and wants to calm them down, he is looking for a larger peace? China is looking for peace across the Taiwan Strait, so perhaps they now have a lot of common understanding on peace.

Trump can better understand China’s position, because for China, to take Taiwan would be very easy. You can see what Iran did by blocking Hormuz. If China were going to block the Taiwan Strait, Taiwan probably would not have to be fought; it could get into real difficulties. I think the separatist movement has no future. Is that also why President Trump recognises reality, and also China’s support for peace in the Middle East and eventually maybe in the Russia-Ukraine war?

Do you think there are many exchanges on these greater global peace and security issues, including peace across the Taiwan Strait, that both China and the U.S. want? Many people interpret Trump’s visit to China as seeking China’s consensus on some of those peace issues and global conflicts. China did make a very strong statement to make sure peace is maintained in the Middle East, and they also talked about Iran. That was in Wang Yi’s statement on the bilateral talks between President Xi and President Trump.

What do you think about the future, and Trump’s position on Taiwan, which is also quite positive, I think? He does not support Taiwan independence, and he does not want to fight 9,500 miles away.

David Shambaugh

Thank you, Henry. I neglected to answer your previous question about the first Trump administration’s National Security Strategy references to China and the second Trump administration’s National Security Strategy references to China. They are very different, obviously.

So what has changed between Trump 1.0 and Trump 2.0? That is, I think, President Trump himself is less encumbered by the bureaucracy, who sees China across the American bureaucracy as a peer competitor. Trump does not have that view of China. Trump does not think in global strategic terms. He is a businessperson. He does have some intuitive sense, though, about great powers, and he understands China in that context. He wants a major commercial relationship with China in both directions: Chinese investment into the United States, and an improved climate for American business here in China, which has had difficulties for a number of years.

It is again an example of this agent-structure paradigm that we call in political science. The agent is the president: he is the leader, and he is imposing his personality and his own views about China, and Russia by the way, onto the bureaucracy. It is affecting American support for Ukraine, and it is affecting the China relationship and Taiwan.

So I will answer your question with regard to Taiwan, but here we have a classic example of a leader who is really imposing his vision. Indeed, in China, the leader is imposing his vision. I am sure the bureaucracy here...I know the bureaucracy here sees the United States in adversarial, competitive terms. But the two leaders met last week, and they have created this new environment. It is not a G2. It never will be a G2. People who argue in that direction, I think, are dreaming. But it is now a managed competitive situation.

Now, with regard to Taiwan, I haven’t been privy to previous discussions at that level about Taiwan between previous American presidents and Chinese leaders, but this time there was some significant. But that is the significant outcome of the summit that I saw. Two things: number one, going into the summit, we were all expecting that the Chinese side would somehow get President Trump to say that he opposes Taiwan independence instead of saying that he does not support Taiwan independence. That did not happen. He did not use that language. There were fears that Trump was going to say he supports the peaceful reunification of Taiwan and China, but Trump did not fall into that trap and did not say that.

When it comes to Taiwan arms sales, Xi Jinping directly asked him—I saw the footage—“Are you going to militarily intervene in Taiwan?” And Trump replied, “I do not discuss that.” That is continuity of the strategic ambiguity principle that the United States has followed with respect to possibly intervening in a Taiwan situation. Possibly. The United States has never, by law or policy, been committed to intervening. Americans have always said it will depend on the situation. The Taiwan Relations Act simply stipulates that the United States shall give Taiwan adequate articles of self-defence, I think is the language.

So, Trump dodged three traps on Taiwan: the “oppose independence” trap, the “peaceful reunification” trap, and the commitment not to sell weapons to Taiwan. That was when he was here in Beijing. Then he gets on the airplane, leaves, goes back, and is very provocative. He says to the media journalists on Air Force One, “I need to talk to the person”—he called him the person; he did not call him the leader or president—“in Taiwan concerning this.”

Well, that is what we are all waiting for now, right? A telephone call between Trump and Lai Ching-te. I do not know if it is going to happen, ladies and gentlemen. If it did, it would be highly provocative. It would violate every principle that the United States has had since 1979. The United States does not talk to the president and the government on Taiwan or the authorities. We do not call it a government; we call it the authorities. But in his first administration, Trump did receive a phone call of congratulations on his election.

So we have to wait to see whether the $14 billion arms sales package is going to go through. I do not know. But last point: if you read Wang Yi’s speech, I think one of the most significant parts of it is that Wang Yi says President Xi really got through to Trump about China’s perspective on the Taiwan question and its historical dimensions. I think this is registered with Trump in a way that it did not previously, and perhaps did not with other previous American presidents. So progress was made in that regard. We will have to wait and stay tuned. But there are three actors here: Beijing, Washington, and Taipei.

Henry Huiyao Wang

I see. I noticed that the Ministry of Foreign Affairs spokesperson mentioned that China opposes any official ties with Taiwan, and any discussion of a possible call. I think the MFA spokesperson strongly condemned that. It is also interesting to hear that, because the U.S. has used up its arsenals and military supplies in the Ukraine war, it may delay or cancel arms sales to Taiwan. That was also in the news.

I do see some positive signs from this visit. At least Trump is saying that they do not want to send soldiers 9,500 miles away. I am sure there will still be another three summits. I hope this is recognised as China’s core interest. Of course, the U.S. has many core interests everywhere now: in Latin America, the Middle East, and Europe. They all need China’s support and understanding, or maybe some consensus there. If we can have a better understanding of each other, maybe we can solve these things much more easily.

Let us come to another subject, perhaps our final two questions. One question is that we saw Trump bring 17 top CEOs, the biggest companies in the U.S., to China. That was really impressive. President Trump said many CEOs still wanted to come, and they only allowed the number ones to come; other senior executives were not able to come. There was a big coverage of those major CEO visits.

The business ties between China and the United States are still pretty strong. To see how eager all those big CEOs were to come shows the interest. The U.S. Chamber of Commerce has also indicated that American interest in China is still very strong. Of course, they also set up two boards, one on trade and one on investment, specifically to increase trade and explore categories of goods that can be exempted from tariffs, as well as investment and what non-sensitive, non-security areas can be invested in.

I think, also, people-to-people exchanges and students are important. Maybe next time, China could announce visa-free entry for Americans. The U.S. is the only G7 country that has not received visa-free treatment. Maybe we can reopen the consulates shut down in Houston and Chengdu. There are many low-hanging fruits that we could pursue.

One of the things we need to get rid of is this over-security issue. Too much of everything is treated as security. There are about 2,000 Chinese companies on the U.S. sanctions list, whereas only about 20 U.S. companies are on the Chinese sanctions list, mainly for selling weapons to Taiwan. With these two boards on trade and investment, maybe we can lower this hype of securitisation, which is everywhere: people-to-people exchanges, students, professors, academics, think tanks, and experts. Over-securitisation is probably something we need to overcome. What do you think we should do to increase business and people-to-people exchanges?

David Shambaugh

Henry, I could not agree with you more. In fact, last night, we were discussing this over dinner. I too feel that the relationship has become over-securitised on both sides towards the other, and there is collateral damage being done to the business community, to the academic community, and to others who should be interacting normally between our two societies.

Both sides have prioritised national security. Xi Jinping has made it the number one national goal. China’s number one national goal is no longer development. 不是发展，现在第一是国家安全, right? That is a change; that is a narrative change. The same is true for the United States. The American side has over-securitised various activities of Chinese entities, not just companies on the entity list in the United States.

I think both sides have to step back and remove these impediments to normal people-to-people interaction. It is going to be very hard to build the framework relationship that they have just stated they are trying to do until they can unwind and dismantle some of these impediments that both sides are putting on. That is the responsibility of each country to define their own national security: What is real national security and what is not necessarily.

So I just wanted to say one last point on the Taiwan arms sales question. Whether the new $14 billion package is approved or not, there is already a six-year backlog of weapons that have not been delivered to Taiwan. So far as the Taiwanese are concerned, it is not the current package they want; it is the six years that are sitting somewhere in the pipeline. The types of weapons that have been moved from Northeast Asia to the Persian Gulf to attack Iran are not the types of weapons that the Taiwanese would be receiving from the United States. Taiwan is not going to get Tomahawk missiles, ladies and gentlemen, and those sorts of munitions.

Henry Huiyao Wang

David, on the securitisation issue, maybe I have somewhat different views from yours. China did not put the U.S. as a strategic rival. China did not sanction thousands of U.S. companies. It was actually the opposite. When China faced the U.S. national security strategy, where China was described as the number one rival, even before Russia, and with 2,000 Chinese companies on American lists, of course China had to put security as a top priority because China felt threatened. There is quite a bit of reinforcement of China’s position, to some extent.

But I do think President Trump’s visit to China is a great way forward. A good example is that under President Trump, we have solved the TikTok issue. It has been solved pretty well and is now operating smoothly.

Maybe we should take advantage of the fact that 17 CEOs came. Next time, when President Xi visits the United States from September 24 to 26—President Trump announced those dates—perhaps China could bring 18 or maybe 20 Chinese CEOs. Let us even have a China-U.S. CEO summit, alternating between the two countries every year. That would be really great. Maybe we can gradually build up some trust.

As you said, we have put a floor at the bottom. I hope now is the bottom, and we can gradually rise up. Of course, we have structural problems. As the two largest economies, there is bound to be some competition, and particular things will happen. But let us not get into conflict. As I discussed with Graham Allison, if we end up in a fight, possibly a nuclear fight, we are going to return to the Stone Age. It would devastate the whole world, and that is not an option. When that option is mutually deterred, the only option left is that we work together. Let us conquer Mars together. Let us fly to the moon together. Let us tackle climate change.

I give high credit to President Trump’s visit this time, and I am sure 2026 will be a good stabiliser for China-U.S. relations.

David Shambaugh

I certainly hope so. And I like very much your suggestion of a CEO summit, which maybe they could do under this new Board of Investment, because we need more Chinese investment into the United States in certain industries. We have national security concerns too, but there is a large variety of sectors in which there are no national security concerns, and Chinese companies should be welcomed into the U.S., and vice versa: American companies here. So we need to improve that dimension.

The last point I would make is: What do we need to be careful about going forward? I would say we need to be more careful about over-securitisation. That is the first thing. I completely agree with you, Henry.

The second thing we need to be careful about is not over-institutionalising the relationship. There is a temptation to go back—this is what governments do: “Let us set up working groups, then we will turn a working group into a vice-ministerial dialogue, turn a vice-ministerial dialogue into a ministerial dialogue, and then we will go up to even higher.” During the Obama administration, there were no fewer than 98, basically 100, bilateral U.S.-China dialogues at the government-to-government level. Trump came in 2017 and cut them to under five. Good. It was a waste of airplane fuel, flying these delegations back and forth, preparations going into these meetings, and then they would produce some communique, perhaps at the end, and then the communique would not be implemented.

Many countries, not just the United States, including European countries and others, developed what they called “dialogue fatigue”. China loves dialogues, and dialogue is intrinsic to diplomacy. But I think there is a temptation, now that we are to put a floor under the relationship, to reconstruct the architecture of dialogues and ministerial interactions. I believe in working groups. Small is beautiful. You do not need this big edifice, because it is never implemented. So I think we have to be careful not to go there.

Henry Huiyao Wang

I have a different view on that, because China is very highly centralised. There is always a top-down approach. Government bureaucrats have to work through these things, and then they can trickle down to business. Different countries may start differently. For the U.S., a smaller government is better and fewer meetings are better. But in China, it is really about building consensus through collectivism, and then trickling down. I still think we need high-level summits, like what happened just last week, and also high-level meetings. I think eventually they will promote better relations. Maybe at the Obama time there were too many, but now there are too few. We need a good balance.

David Shambaugh

Wang Yi identified eight specific areas in his press conference that the two sides agreed to. This is critical. The American administration, after the summit, did not say anything about those eight areas. It is the Chinese side that said eight specific areas. So I would imagine and hope that they can establish working groups in all eight areas: public health, and all eight. That is the first step. My suggestion is that I do not want it to become too large. I want it to be efficient, focused, and professional.

This brings us back to another article in the book Pioneering Through Hardship: An Account of the Decision-Making and Construction of Landmark Engineering Projects (筚路蓝缕：世纪工程决策建设记述), which I introduced in my last piece, namely “The Origins and Development of the National Strategic Petroleum Reserve.”国家战略石油储备的发展

Author Zhang Guobao served as Vice Chairman of the National Development and Reform Commission (NDRC) from 2003 to 2011, was the first head of the National Energy Administration, and also led the West–East Gas Pipeline project. It was during his tenure that China built its national strategic petroleum reserve system from scratch and defined its operating model. Also, big thanks to my friend Mao Keji for the book recommendation. Every conversation with him is a lesson. from renowned newsletter did an interview with him last year, and I highly recommend it.

The Aoshan base in Zhoushan that Li Qiang inspected is in fact one of the Phase I national strategic petroleum reserve bases whose construction Zhang personally oversaw more than two decades ago. As an old Chinese saying goes, one must know not only what something is, but also why it is so. 知其然，知其所以然 I would like to take this opportunity to introduce Zhang’s account of that decision-making process to my readers.

The article opens with the first oil crisis of 1973, traces the birth and maturation of emergency petroleum reserve systems in Western countries, and then turns to the State Council’s formal decision in late 2002 to launch construction of national petroleum reserve bases. The most valuable parts of the piece are Zhang’s detailed accounts of three core decisions, namely siting, storage format, and operating model, along with lesser-known episodes such as seizing low oil prices during the 2008 financial crisis to fill the reserves, the push to domestically produce the high-strength steel plates needed for storage tanks, and attempts to expand reserves through idle tankers and futures operations.

My understanding is that the article tells several characteristic features of the Chinese decision-making model. The first is a flexible blend of government leadership and enterprise coordination. For the operating model, Zhang proposed a three-tier management structure. At the top sits the Petroleum Reserve Office of the National Energy Administration, responsible for policy and planning. In the middle is the National Petroleum Reserve Center, a public-service institution responsible for project implementation and day-to-day management. At the operational level, the actual work of construction, management, stockpiling, and release is contracted out to nearby state-owned oil companies (Sinochem, Sinopec, and CNPC). This setup preserves the government’s ultimate authority over strategic reserves, while drawing on the human capital, industry resources, and market responsiveness of the central state-owned enterprises, and avoids the fiscal cost of building yet another large bureaucracy from scratch.

The second is what the Chinese call “crossing the river by feeling the stones,” a kind of pragmatism that learns from Western experience without copying it wholesale. Reading Zhang’s account, one can clearly sense the influence of his study of the U.S. Strategic Petroleum Reserve. Yet when it came to institutional design, he placed greater weight on China’s own industrial capacity and market conditions. At the time, China lacked a mature ecosystem of oil companies and oilfield service providers; the market was overwhelmingly concentrated in the “three oil majors,”三桶油 and a fully marketized, U.S.-style competitive bidding system would have been neither meaningful nor feasible. At the same time, Zhang opposed the NDRC, a government agency, sending officials directly to run the bases. The three-tier management system emerged as the compromise. This habit of referencing without replicating is itself a fairly typical pattern in Chinese policymaking.

The third is gradualism, testing first and then scaling up. Phase one aimed simply to solve the problem of having a reserve at all. All sites were located along the eastern coast, close to demand centers, near large refineries, and convenient for seaborne imports, and they used the relatively mature aboveground storage tank format. Only after experience had accumulated and engineering capacity had improved, did Phase II gradually introduce the technically more demanding underground water-sealed rock cavern storage and extend further inland. This step-by-step quality is itself a distinctive part of the Chinese approach.

That said, Zhang’s memoir does not omit the shortcomings of the existing decision-making mechanism, and he repeatedly recounts how he tried to “work the edges” within the system to circumvent procedural inefficiencies. Funding for the petroleum reserve comes from the central budget. Each year’s stockpiling plan must first be drafted by the Reserve Office, then approved by the Ministry of Finance and the State Council before it can be carried out. The cycle is long, and opportunities to buy at price troughs are often missed. During periods of depressed oil prices, Zhang considered leasing idle tankers as floating storage and partnering with privately owned tank farms to expand capacity, but the idea fell through because the Ministry of Finance refused to pay rent to private operators. On another occasion, when prices fell below the $40-something range, Zhang told state-owned enterprise executives at a meeting to start stockpiling first and complete the paperwork later, saying he would take personal responsibility if anything went wrong. The companies, however, did not dare move, and the window closed. His candor about this tension between compliance and efficiency leaves a great impression.

Subscribe now

Below is the English version of the article I made

The Decision-Making Process Behind China’s Strategic Petroleum Reserve

In 2015, international oil prices declined continuously. By late January 2016, U.S. West Texas Intermediate (WTI) crude oil futures had fallen below $30 per barrel, a drop of more than 60%. Currently, international oil prices hover around $50 per barrel and remain in a slump. Meanwhile, in 2016, China imported 380 million tons of crude oil, reaching a record level with a growth of 13.56%.

Faced with the sharp drop in international oil prices, everyone agrees that China’s earlier decision to establish strategic petroleum reserve bases during the period of high oil prices was an extremely wise and timely move. After the reserve bases were completed, they coincided with the drop in oil prices, allowing for bargain purchases that significantly reduced the cost of stockpiling oil, providing strong energy security for China’s future economic development.

As a participant in this decision-making process, looking back, I deeply feel that the decision was extremely correct and timely. However, the process was accompanied by repeated comparative analyses, demonstrations, overcoming difficulties, and accumulating experience—it was not as simple or easy as it might seem. This decision-making process fully reflects the strong sense of responsibility of China’s energy workers in ensuring national energy security, as well as their persistent efforts.

International Background

Internationally, strategic petroleum reserves were born out of the first oil crisis after World War II, strengthened during the second oil crisis, and continuously matured and improved through subsequent oil price fluctuations. During the first oil crisis of 1973-1974, the Organization of Petroleum Exporting Countries (OPEC) controlled production, causing crude oil prices to rise from $3 to $11 per barrel. This dealt a heavy blow to Western economies heavily dependent on oil imports, making developed Western nations realize that oil supply was an “Achilles’ heel” of their economies. Therefore, in 1974, OECD countries jointly established the International Energy Agency (IEA), requiring member countries to maintain reserves of at least 60 days of import volume to cope with oil crises. This is known as emergency oil reserves. Oil reserves include two forms: government reserves and corporate reserves, and member countries should provide mutual reserve support when necessary. The government reserves are also known as strategic petroleum reserves.

During the second oil crisis in the 1980s, primarily triggered by the Iranian Revolution, oil prices rose from $13 to $43 per barrel, further highlighting the vulnerability of oil-dependent economies in the face of oil price crises. The IEA then required member countries to increase oil reserves to more than 90 days of net imports. The oil reserve system was gradually improved and expanded. Since then, over 30 years have passed, and whenever international oil supply faces risks, IEA member countries mobilize their oil reserves. For example, the U.S. government released 33 million barrels of reserve crude oil to the market through direct sales during the 1991 Gulf War. The United States also once asked China, as another oil-importing country, to coordinate with the U.S. in releasing crude oil reserves and stabilizing international oil prices.

Through emergency oil reserves, Western countries have: first, effectively weakened oil-producing countries’ ability to use oil as a weapon to deter Western countries, preventing artificial supply shocks from occurring or occurring frequently; second, in the event of an actual supply crisis, mitigated the crisis risks by releasing crude oil reserves, minimizing the impact of oil supply shocks, and ensuring their own economic and political stability. It can be said that emergency oil reserves have become an important and effective energy protection measure and economic weapon for Western countries.

China’s Background

When Western countries established emergency oil reserves in the 1980s, some Chinese comrades who closely followed international developments proposed that China should also establish a strategic petroleum reserve. However, at that time, China’s oil supply still exceeded demand—China was an oil-exporting country—and there was no strong sense of crisis at the national level regarding oil supply shortages or interruptions. It was difficult to predict at the time that China’s future oil demand would grow rapidly and substantially, that China would become a net oil importer with external dependence exceeding 60%, and that it would become one of the world’s largest oil importers. Therefore, this proposal did not receive particular attention and was not implemented.

The year 1993 marked a major watershed in China’s energy supply and demand. Due to rapid economic development, crude oil imports increased sharply. That year, China transformed from a net crude oil exporter to a net crude oil importer, ending 30 years of oil self-sufficiency and slight surplus exports since the discovery of the Daqing Oilfield. Since then, crude oil imports have continuously reached new milestones. In 2004, they exceeded 100 million tons for the first time; in 2009, they exceeded 200 million tons; in 2014, they exceeded 300 million tons. In 2015, China’s oil import dependence reached 60.6%, and in 2016 it further rose to 65.4%. It was against this dramatically changing energy landscape that the necessity and urgency of establishing China’s strategic petroleum reserve became increasingly prominent. The research community and decision-makers gradually unified their views.

As China’s oil consumption and imports increased year by year, the Party Central Committee and the State Council, from the overall perspective of modernization and the need to safeguard national energy security, repeatedly issued important directives on establishing a national strategic petroleum reserve and ensuring national energy and economic security.

On March 17, 1996, the Fourth Session of the Eighth National People’s Congress approved the “Outline of the Ninth Five-Year Plan and the Long-Term Goals to 2010 for National Economic and Social Development of the People’s Republic of China,” which proposed “strengthening oil reserves.” In November 2000, at the Central Economic Work Conference, then-General Secretary Jiang Zemin explicitly required that projects related to the overall situation, such as establishing a national strategic petroleum reserve, be promptly studied and implemented as soon as possible. On March 20, 2002, the State Planning Commission submitted “Several Points on Increasing Crude Oil Reserve Facilities Construction” to the State Council. Then-Premier Zhu Rongji wrote on the report: “Please have the Planning Commission further demonstrate this, and submit it to the Premier’s Office Meeting for review and approval.” At the end of 2002, the Premier’s Office Meeting of the State Council heard and reviewed and approved “The State Planning Commission’s Request for Instructions on Establishing the National Petroleum Reserve Implementation Plan” (Ji Zonghe [2002] No. 2082), marking the official launch of China’s national petroleum reserve base construction.

Key Decisions in Construction

The construction of strategic petroleum reserve bases was put on the agenda. But where exactly should they be built? What management model should be adopted? Should aboveground oil depots or underground reserve depots be built? These questions, while seemingly very specific and minor, would greatly affect the construction process and work efficiency if not properly resolved. These decisions all underwent very detailed demonstration and research.

The First Question: Where to Start Construction?

When news spread that the state was preparing to build a strategic petroleum reserve, various regions became very enthusiastic, seeing it as an important opportunity to stimulate the local economy and the development of local petrochemical industries. However, due to China’s vast territory, varying levels of economic development, and existing petrochemical production capacities, plus the fact that this was an entirely new undertaking, pilot projects were necessary to accumulate experience before full-scale implementation. To this end, we invited the China International Engineering Consulting Corporation to organize experts to evaluate the site selection and other construction conditions, project investment, and safety production facilities for the national petroleum reserve base project. They proposed that “the establishment of a national petroleum reserve base should follow the principles of unified planning, rational layout, standardized management, and step-by-step progress, fully relying on and utilizing existing facilities.”

The first phase was to address the question of “to have or not to have,” choosing as soon as possible to build in the eastern coastal areas. The eastern region had large oil demand, more existing refineries, and oil reserves could serve refineries nearby. Meanwhile, the eastern coastal area had the advantage of being convenient for importing crude oil by sea.

In 2004, except for a small amount of crude oil transported across borders by railway from Kazakhstan and Russia, the vast majority of China’s imported crude oil was transported by sea.

In the end, four reserve projects were selected for the first phase, including the Zhoushan Aoshan, Ningbo Zhenhai, Qingdao Huangdao, and Dalian bases. All four locations were in the eastern coastal area, economically developed, with strong demand, and all had large refineries nearby. It was also determined at the time that subsequent second and third phase reserve projects could be built in suitable western regions to make the layout more rational and secure. Later, multiple locations in the southwest and northwest enthusiastically applied to build reserve bases. The second phase project arranged for several western petroleum reserve bases.

The Zhoushan Aoshan Port has an annual oil throughput of nearly 27 million tons. Aoshan Island is currently the largest petroleum reserve and throughput base in China.

The Second Question: What Management System to Adopt?

The U.S. strategic petroleum reserve management model is divided into two levels. At the management level, the Strategic Petroleum Reserve Office of the U.S. Department of Energy, located in Washington, D.C., is responsible for reserve policies and planning; the Project Management Office located in New Orleans is responsible for the implementation, operation, and management of specific projects. At the operational level, a market-based mechanism is adopted, with oil companies and base management companies determined through market bidding. Oil companies are responsible for procurement and release of reserves, while base management companies are responsible for the daily operation, maintenance, and security protection of reserve bases.

At the time, some comrades in the National Development and Reform Commission and the Ministry of Finance suggested partially imitating the U.S. model. The management level would be the same as in the U.S., but at the operational level, they proposed establishing specialized base management agencies responsible for base management as well as the procurement and release of crude oil at the bases. In other words, the Petroleum Reserve Office would manage everything from top to bottom—personnel, finances, materials, production, supply, and sales—forming a top-down management system. Other comrades suggested fully imitating the U.S. model.

But at the time, I believed that both approaches had flaws. If specialized base management agencies were established and personnel recruited, the state would have to pay various expenses for these agency staff, adding extra cost burdens to reserve work. Moreover, since talent and resources were concentrated in a few large state-owned oil and petrochemical enterprises at the time, it would be impossible to immediately recruit so many workers familiar with oil storage, procurement, and sales, which would inevitably affect reserve work efficiency. An office in Beijing might appear to have authority, but remotely directing base work could not effectively seize market opportunities for stockpiling oil, and in the event of safety accidents, responsibilities would be difficult to define. Fully copying the U.S. model was not feasible given China’s national conditions at the time. The U.S. oil market is fully market-oriented, with a large number of specialized oil companies and service companies that can quickly determine oil and management companies through bidding. But China’s oil market did not have the same conditions as the U.S., and this could not be achieved in the short term.

Both models had shortcomings. I believed that the U.S. model’s design at the management level was good. The main issue was at the operational level. To save operating costs and improve work efficiency, the U.S. model of determining the operational level through market bidding should be adopted. However, China’s oil was mainly concentrated in three major oil companies, and large refineries belonged to these three oil companies. Looking at it from another angle, China’s market structure is dominated by state-owned enterprises, and the interests of state-owned enterprises should be consistent with national interests. So we could directly delegate base management and operation as well as crude oil procurement and release responsibilities to oil companies near the bases. This would not only save various operational costs for the state but also fully utilize the talent and resource advantages of state-owned enterprises, improving work efficiency. At the same time, this allowed state-owned enterprises to obtain more information, hardware, and funds for their own operations through managing oil reserves. Since the construction and operational costs of oil reserves were funded by the national finances, this approach was also more acceptable to the Ministry of Finance.

At relevant meetings, I proposed establishing a three-level management model for petroleum reserves with Chinese characteristics, and everyone expressed support. A Petroleum Reserve Office would be established under the National Energy Administration as a government management agency responsible for top-level reserve policies and planning, representing the government in deciding the scale of stockpiling and the use of petroleum reserves. An institution similar to the U.S. Project Management Office would serve as a middle-tier public institution specifically responsible for project implementation and operational management. The operational level would be entrusted to nearby state-owned enterprises, which would be responsible for constructing and managing the corresponding petroleum reserve bases and bearing safety responsibilities. Crude oil reserve release would be operated by the enterprises according to the national plan prepared by the Petroleum Reserve Office.

Following this approach, in 2003, the Energy Bureau of the former National Development and Reform Commission added the sign of the National Petroleum Reserve Office, which later, with institutional reforms, was adjusted to become the current Reserve Office of the National Energy Administration, serving as the head of the petroleum reserve system. On December 18, 2007, the National Petroleum Reserve Center was officially established as a public institution under the National Development and Reform Commission, later adjusted to become a public institution under the National Energy Administration. As the middle tier of China’s petroleum reserve management system, it exercises investor rights, is responsible for the construction and management of national petroleum reserve bases, undertakes strategic petroleum reserve stockpiling, rotation, and utilization tasks, and monitors supply and demand changes in domestic and international oil markets. The operational level consists of reserve base companies funded by the state and entrusted to state-owned enterprises for management. Of the four first-phase bases, Zhoushan base was entrusted to Sinochem Corporation for management; Zhenhai and Huangdao bases were entrusted to Sinopec’s Zhenhai Petrochemical Company and Qingdao Refinery for management; Dalian base was entrusted to PetroChina’s Dalian Refinery for management. This saved on additional personnel allocation and also increased the daily operational management and safety responsibilities of the enterprises.

The Third Question: Aboveground Oil Depots or Underground Reserve Depots?

Since the United States has large salt domes concentrated in Louisiana and Texas near the Gulf of Mexico, close to the petrochemical industrial belt, most U.S. strategic petroleum reserve facilities are underground salt cavern oil storage facilities. Constructing such caverns is very easy—water is pumped into the salt dome to dissolve the rock salt, then the brine is extracted, leaving large underground cavities with good sealing properties. Construction and maintenance costs are very low. Countries without such natural conditions can only adopt the approach of building large aboveground oil tanks. There is also a type of underground water-sealed rock cavern, which refers to artificially excavating caverns of certain shapes and volumes in rock formations below the groundwater table for oil storage. This has the advantages of small land occupation, low construction and maintenance costs, and being safe and reliable. Being below the groundwater table ensures that the water pressure in the fissures is higher than the oil pressure inside the cavern, preventing oil leakage. Many regions in China have the geological conditions to build underground water-sealed rock caverns.

There was also debate about which approach to take. Some comrades believed that strategic petroleum reserves were related to national energy security and should imitate the U.S. model by building underground reserve facilities to maximize reserve safety. But eastern China did not have underground salt domes like those in the U.S. Building caverns required directly excavating in underground rock, with construction costs far exceeding aboveground tank construction. It was difficult to balance cost and safety. We held multiple meetings to discuss this. We comprehensively analyzed the main factors affecting reserve safety, including war factors and terrorist attack factors. These were problems that all countries generally faced, not unique to China. Therefore, we should not refrain from building aboveground oil depots. Future reserve facilities should combine aboveground oil depots with underground reserve facilities.

For the first phase, we would first address the question of “to have or not to have.” Everyone thought this made sense. Therefore, all four first-phase bases adopted the aboveground tank approach. Of course, considering that we had built underground water-sealed rock caverns for LPG in places like Yantai and Shantou, in order to accumulate experience with underground reserve facilities, we also considered adapting to local conditions and planning to build large underground water-sealed rock caverns in places like Qingdao Huangdao, where aboveground space was limited and underground geological structures were favorable. Using underground salt domes for oil storage was not adopted due to issues such as existing underground salt mines being far from the coast, insufficient salt layer thickness, and difficulty in brine treatment.

Construction Begins

After these issues were resolved, in March 2004, the National Development and Reform Commission held the launch meeting for the first-phase project construction of the national petroleum reserve bases. At the meeting, I delivered a mobilization report on advancing the first-phase project construction and signed project construction responsibility agreements with leaders of several group companies. The Zhenhai base was the first to begin construction, opening the curtain on China’s establishment of national petroleum reserves.

However, the first-phase project construction did not proceed entirely smoothly. The project encountered problems with the trial production and supply of domestic high-strength steel plates for large storage tanks, while the prices of imported steel plates multiplied. For this reason, I decided to organize relevant departments to undertake the localization of high-strength steel plates for large storage tanks. Through joint efforts from various parties, high-strength steel plates produced by Shanghai Baosteel Group, Wuhan Iron and Steel Group, and other companies successively passed certification by the National Pressure Vessel Standardization Technical Committee and welding process experiments by construction units. They were successfully promoted and used at reserve bases, breaking the monopoly of Japanese steel companies in China’s high-strength steel plate market, saving the country a large amount of construction funds, and improving the manufacturing level and competitiveness of domestic steel enterprises.

Through the joint efforts of project builders, the Zhenhai, Huangdao, Dalian, and Zhoushan bases were completed and put into operation in September 2006, December 2007, November 2008, and December 2008, respectively.

The completion of the first-phase project coincided with the outbreak of the 2008 global financial crisis, with international oil prices dropping from $150 per barrel to $50 per barrel. The four bases seized the opportunity to stockpile large amounts of cheap crude oil, completing oil injection in the first half of 2009. The average stockpiling cost was only about $56 per barrel. The base construction and crude oil stockpiling were a great success. I also wrote a special report on this to submit to State Council leaders.

Based on the successful experience of first-phase project construction, we organized and began compiling petroleum reserve plans. In 2008, the State Council approved the “National Petroleum Reserve Medium and Long-term Plan,” planning to complete three phases of strategic petroleum reserve base construction by 2020. Starting in 2009, the second-phase reserve bases were launched, with new reserve bases including Zhoushan (Phase II) and Huangdao (Phase II). The second-phase reserve bases included western regions and underground reserve facilities, making the reserve layout more comprehensive and the reserve methods more rational. The construction of national petroleum reserve bases entered a phase of normal progress.

Other Episodes Related to Petroleum Reserves

During this period, several other events related to petroleum reserves occurred.

In October 2008, I went to the Zhoushan base to inspect construction progress, accompanied by Comrade Zhong Ren from Sinochem Corporation, who was in charge of crude oil futures business. Comrade Zhong Ren was an expert in crude oil futures operations and had obtained substantial profits for Sinochem through futures hedging. He had been honored as a National Model Worker. During the inspection, he introduced crude oil futures to me, and I learned a lot about hedging. One evening after dinner, while taking a walk, I was chatting with my secretary.

I said: International oil prices are currently falling, our reserve facilities are nearly full, and we have no way to buy more oil. To seize the opportunity of falling oil prices, we could lock in some cheap crude oil by purchasing futures. Even if futures prices continue to fall in the short term, they will certainly rise again in the long term. As long as our purpose of buying futures is not speculation but to obtain physical crude oil in the future, we will definitely buy cheap crude oil in the long term. I judge that oil prices are currently at a low level, and taking the opportunity to buy futures should be cost-effective. Even if price fluctuations occur, as long as we are not speculating, the risks are controllable.

In early spring 2009, I went to inspect the construction of the Dalian petroleum reserve base. Comrades from the Dalian shipbuilding industry told me that due to the international financial crisis, many shipowners had tight cash flows and could not pay shipbuilding fees, resulting in many new ships, including oil tankers, being stranded at shipyards. At the time, there were reports that some European countries were using idle oil tankers to stockpile crude oil at low prices. I thought at the time that we might as well lease these semi-finished oil tankers that shipowners did not want. Without installing power or other equipment, we could directly tow the sealed hulls to nearshore waters as floating oil storage bases, which would solve the current problem of insufficient storage capacity—a multi-purpose solution.

I shared this idea with the comrades from the National Petroleum Reserve Center who accompanied me, and everyone thought it was a good approach. Later, the National Petroleum Reserve Center cooperated with China Shipbuilding Heavy Industry Group to research and propose a plan to directly manufacture Very Large Crude Carrier (VLCC) modules at the shipyard and place them on hard bedrock tidal flats in a sit-on-bottom manner, building large petroleum reserve facilities in a short time.

When oil prices were low, I hoped to take advantage of low prices to increase stockpiling. At that time, most reserve bases had not been completed, and we had considered using private oil depots, but private oil depots were small in scale, and their port locations could not accommodate large oil tankers. More importantly, the Ministry of Finance did not agree to pay rent to private oil depots, so this did not materialize.

I wrote specific reports to the State Council on these ideas. However, possibly due to the complexity of the operations, the Ministry of Finance had different opinions, especially regarding futures operations, which everyone still had doubts about. So they were not actually implemented. Looking back now, it is very regrettable. But I think these ideas can serve as a vision for future diversified petroleum reserves. Gratifyingly, there are now private enterprises building 3 million cubic meters of private reserve facilities in Yangpu, Hainan, pioneering private enterprise participation in stockpiling.

Due to our system, petroleum reserve funds are provided by the Ministry of Finance. The annual stockpiling amount is first proposed by the Reserve Office, then sent to the Ministry of Finance for joint signing. After the Ministry of Finance agrees, it is jointly submitted with the National Development and Reform Commission to the State Council for approval before implementation. So this decision-making process is very slow, often missing the best opportunities to stockpile when oil prices are low.

Once, when oil prices fell to over $40 per barrel, I urgently convened a meeting. I believed that oil prices had basically fallen to the bottom and we should stockpile quickly. I asked PetroChina, Sinopec, and Sinochem to act quickly while initiating the procedures for joint signing with the Ministry of Finance and submission to the State Council. I knew that international oil prices change rapidly, and by the time approval came through, the moment would have passed. So at the meeting, I said: After everyone returns, act immediately. Don’t wait for approval to come down before acting. If problems arise, I will take responsibility. But no one dared. PetroChina bought a small amount, but later, seeing no approval document, sold it off.

Another incident: On July 16, 2010, a crude oil tank fire accident occurred at PetroChina’s Dalian commercial reserve depot. I happened to be on a business trip to Dalian. Because I did not know in advance which oil depot was on fire, I thought it was the Dalian reserve base. I quickly rushed to Dalian’s Xingang Port to participate in the firefighting operations organized by the on-site emergency command. Only later did I learn that it was not the Dalian reserve base but a commercial reserve depot of an enterprise. Afterward, an accompanying staff member asked me whether insurance had been taken out. Inspired by this, I thought about the insurance issue for state-stored petroleum. Upon returning to Beijing, I immediately convened a meeting. At the time, the national petroleum reserve bases were indeed not insured, and there had even been lightning strikes.

The Reserve Office and Reserve Center then negotiated with the Ministry of Finance. With the support and leadership of the Ministry of Finance, starting in 2011, the National Petroleum Reserve Center selected several insurance companies through bidding and insured all assets of the four first-phase bases, also building lightning protection devices.

Underground Water-Sealed Rock Caverns

By the time China’s first large underground water-sealed rock cavern was completed, I had already retired. I took advantage of the time before the cavern was sealed to go deep into the underground cavern for a detailed inspection. Underground water-sealed rock cavern oil storage is quite common in Europe, especially in Sweden where it has a history of sixty to seventy years, from military to civilian use. They had also wanted to transfer technology to China, but later we had mastered the technology ourselves and could basically build it independently. Through the inspection, I corrected the previous misconception that underground water-sealed rock caverns were expensive to build—in fact, construction costs are even lower than aboveground oil storage facilities, with less land occupation and lower operating costs. Only the construction period is longer. China has very strong tunnel and culvert construction teams in hydropower and railway construction, and we should strive to build more underground water-sealed rock caverns where conditions permit.

I later reported the situation of the first large underground water-sealed rock cavern to relevant departments through the National Development and Reform Commission’s “Economic Situation and Suggestions,” which received attention and was also prepared for application to refined oil reserves. In subsequent construction, multiple underground water-sealed rock caverns were planned, some of which have already been completed.

The Aoshan Base

The reserve base located on Aoshan Island in the Zhoushan Islands is the only base managed by an entity other than the three major oil companies—it is managed by Sinochem Group, because Sinochem Group originally had its own enterprise reserve depot at Aoshan. After the Aoshan base was completed, it was spectacular. Sinochem Group organized retired cadres to visit, and these old comrades were extremely excited about the country’s construction achievements when they saw each 100,000 cubic meter large oil storage tank that had been built.

When General Secretary Xi Jinping was serving as a leader in Zhejiang Province, he was very concerned about the construction of the Aoshan petroleum reserve base and visited the base on the island. After becoming General Secretary, when inspecting Zhejiang, he made a special trip to inspect the Aoshan base and gave important instructions.

When a central leader inspected the Aoshan base, base leaders reported that there were still a small number of residents on the island and hoped the state would fund the relocation of all residents on the island, turning the entire island into a reserve base for greater safety. After receiving the opinion forwarded by this central leader, I coordinated with the Investment Department of the National Development and Reform Commission to allocate 300 million yuan, turning the entire Aoshan Island into a reserve base.

After the Aoshan base was completed, it was spectacular. Sinochem Group once organized retired cadres to visit, and these old comrades were extremely excited about the country’s construction achievements when they saw each 100,000 cubic meter large oil storage tank.

When I saw the rows upon rows of huge white storage tanks, I was also very moved and composed a poem to the tune of “Xi Jiang Yue”:

Xi Jiang Yue · Zhoushan Aoshan Island西江月·舟山岙山岛
Commemorating the Completion of the National Petroleum Reserve Base建成国家石油储备基地

The East China oil pipeline begins at Aoshan, passes through Cezi Island, crosses the Xihoumen Strait shipping channel, lands at Jintang Island, and goes along the Yangtze River to Wuhan.

华东输油管道起自岙山，经册子岛，穿过西堠门海峡航道，经金塘岛登陆，沿长江直到武汉。
Five years of planning and construction, finally bearing fruit.

五年筹划建设，终成正果。

Putuo, the Buddhist realm of sea and sky,

海天佛国普陀，散落岛屿星罗，
Islands scattered like stars,
Aoshan suddenly displays white encampments,
Erecting giant tanks one after another.

岙山惊现白营盘，筑起巨罐座座。

Looking like masts joined to the clouds,
Yet it is national reserve crude oil.

观似连云樯橹，却是国储原油。
Sea pipelines pass through Xihoumen,
Directly supplying Jiangsu, Anhui, Hunan, and Hubei.

海上管道穿西堠，直输苏皖湘鄂。

In the past, many petrochemical plants were laid out along the Yangtze River, such as Yangzi Petrochemical, Anqing Petrochemical, Jiujiang Petrochemical, Wuhan Petrochemical, Yueyang Petrochemical, etc. The crude oil they needed was transported upstream by Changjiang Shipping Company’s vessels. After the Aoshan base and the Yangtze River pipeline were completed, the crude oil needed by these petrochemical enterprises along the river was mainly switched to pipeline transportation, saving transportation costs. But a problem arose: the operations of Changjiang Shipping Company were affected. For this reason, I conducted special research and approved Changjiang Shipping to develop ocean transportation business, and asked Sinopec to give Changjiang Shipping some refined oil transportation business.

The completion of the first-phase national petroleum reserve bases coincided with the oil price drop caused by the 2008 international financial crisis. When the second-phase national petroleum reserve bases were successively completed, they coincided with the oil price drop caused by the 2015 global economic slowdown. China’s strategic petroleum reserve has accumulated very rich and successful experience in site selection and layout, operational models, and timing.

Facing the rapidly changing international economy and oil market, as long as we continue to learn from international experience, realistically combine it with China’s national conditions, remain humble and prudent, and pioneer innovation, we will surely grasp the laws of economic development and international oil price changes, build a good strategic petroleum reserve, and provide solid energy security for the country’s economic construction.

I’m probably the first one to introduce this book to the English-speaking community. What makes this memoir unusual is that it mentions the thinking and bargaining details behind major policy decisions: from the initial vetting of proposals, to disagreements between ministries, to balancing local demands against central objectives, and finally to the top leadership’s call. In recounting the deliberations behind the West-East Gas Pipeline, he lays out in detail how the route, gas sources, and natural gas pricing mechanism were determined. Discussing the development of Chinese wind power, he does not shy away from the large-scale “wind curtailment”（弃风） that resulted from inadequate transmission line buildout during the buildout phase. He also writes in detail about the negotiations over the China-Russia crude oil pipeline, including his complaints about the Russian negotiating style and how the two sides maneuvered against each other. In my view, for anyone trying to understand the logic and considerations behind Chinese government decision-making, this book has an irreplaceable value.

Subscribe now

Zhang served as Vice Chairman of the NDRC from 2003 to 2011, holding posts as the first Director of the National Energy Administration and as head of the leading group for the West-East Gas Pipeline project. He oversaw the formulation of industrial plans and policies for shipbuilding, automobiles, steel, and a host of other key sectors. More importantly, his tenure coincided with the pivotal eight-year window in which Chinese new energy shifted from a relative fringe option to a national strategy.

During his time in office, the Renewable Energy Law was enacted in 2005, providing policy support for wind and solar; installed wind capacity leapt from under 600 MW in 2003 to over 60 GW in 2011, making China the world’s largest wind power market virtually overnight. Gigawatt-scale wind bases in Jiuquan (Gansu), Inner Mongolia, Hebei, and elsewhere broke ground in succession. The solar PV industry also gradually pivoted from export dependence toward the domestic market. At the same time, a slate of projects that define today’s energy landscape—ultra-high voltage transmission, third-generation nuclear power (including the introduction of the AP1000), and large hydropower projects such as Xiluodu溪洛渡 and Xiangjiaba向家坝 were also given the green light during this period. Because of the singular depth of this experience, I'm planning to spend some time, through a series of newsletters, sharing pieces of his memoirs.

It is worth noting that this book is considered essential reading within China's Industrial Party (工业党) circles, a tech-savvy online group that argues that engineering and scientific capability are the primary drivers of national power. If Illumine Lingao临高启明 represents the cultural archetype of this current of thought, then Zhang Guobao's memoirs is a rare top-down perspective; it's a real-world affirmation of how the Industrial Party ethos manifests at the level of national governance.

I chose "A Fifteen-Year Negotiation: The China-Russia Crude Oil Pipeline" as the opening piece of this newsletter series. Published in 2018, the article is Zhang’s detailed recollection of the 15-year history of pipeline negotiations stretching back to the 1990s. It's a window into how geopolitics, diplomatic relations, and commercial bargaining shaped the cross-border infrastructure decision. He also shares many rarely-seen negotiating details, like dozens of hours of continuous talks with the Russian Deputy Energy Minister, only to share cigarettes afterward and joke about U.S. Energy Secretary Bodman. A last-minute change of stance from the Russian side just before signing, leaving the Chinese and Russian Vice Premiers waiting at the Great Hall of the People for three hours while their teams hashed things out on the spot. After the talks finally concluded, Zhang even wrote a classical-style poem to commemorate the grueling experience. I believe these are the kinds of details that no official readout could ever convey.

A Fifteen-Year Negotiation: The Inside Story of the China-Russia Crude Oil Pipeline

Why Build the China-Russia Crude Oil Pipeline?

China was once an oil-poor country. The gasoline, kerosene, and diesel it used almost entirely depended on imports—hence the term “foreign oil.” In the late 1950s and early 1960s, the Daqing Oilfield was discovered and developed, and in 1963 China achieved oil self-sufficiency, with some surplus exported for foreign exchange. But as the economy grew and living standards rose, demand surged. Self-sufficiency lasted only 30 years: by 1993, imports exceeded exports and China once again became a net crude oil importer—a trend that has grown year after year. By 2016, China imported 380 million tons of crude while producing 200 million tons domestically, putting external dependency above 60%.

In the 1990s, extraction at Daqing gradually shifted from natural flow and pumping to water injection, with water injection accounting for an ever-larger share. Tertiary recovery techniques were later developed, using chemical agents with detergent-like properties to wash oil out of rock fissures. The petroleum sector recognized that Daqing was moving past its peak into decline. Yet Daqing had become the main crude source for China’s major refineries: its oil flowed through pipelines to refineries in Jilin, Liaoning, and other areas. As Daqing output fell, those refineries faced supply problems. So from the 1990s onward, the petroleum sector began contacting the Russian side, exploring a pipeline from Siberian oilfields to Daqing to compensate for Daqing’s declining output.

Yukos: The Initial Private-Sector Partner

The original idea was to build a pipeline from Russia’s Sakha and Chayanda fields to Daqing, but negotiations with Russia were extremely difficult and made no substantive progress.

In 1994, CNPC made contact with Yukos, a private Russian oil company. Yukos was more enthusiastic than Russia’s state oil company about cooperating with CNPC on a China-Russia oil pipeline. After the Soviet collapse, many private enterprises emerged; one of the larger ones was Yukos. Its chairman, Mikhail Khodorkovsky, had been a Komsomol secretary in the Soviet era. After the USSR’s dissolution, he went into business and founded both a company and a bank; affiliated entities acquired Yukos shares during the privatization wave and gradually came to hold 90% of Yukos. The company was a joint-stock firm whose raised capital included shares held by American citizens.

Yukos proposed the “Anda Line”（安大线）—a pipeline from Russia’s Angarsk oilfield to Daqing. It would start at Angarsk in Irkutsk Oblast, head south into the Republic of Buryatia, skirt Lake Baikal, run east through Chita Oblast, enter China, and reach Daqing. This fit CNPC’s thinking, and so Yukos remained the principal negotiating counterpart.

In 1996, the two sides completed a pre-feasibility study for the Anda Line. But by 1999, when I took office as Vice Chairman of the State Development Planning Commission, no result had been reached. In September 2001, CNPC, Yukos, and Russia’s Transneft pipeline company signed a general agreement to conduct a feasibility study on the Anda Line—but the Russian government’s views seemed inconsistent with Yukos’s, and no substantive progress was made.

Zhu Rongji’s Idea: Trade $1.4 Billion in Cash for the Tianwan Nuclear Plant in Exchange for Russian Approval of the Pipeline

After I assumed my deputy-director post and took charge of energy and raw materials, I began participating directly in negotiations on Russian crude imports and pipeline construction. On one occasion, then-CNPC General Manager Ma Fucai马富才 returned from Russia and reported to Premier Zhu Rongji that the Russian side had offered the following: if the $1.4 billion in barter funds that Russia was using to supply equipment to China’s Tianwan Nuclear Power Plant in Lianyungang, Jiangsu could instead be paid in cash, Russia would agree to build the China-Russia crude oil pipeline.

The Tianwan plant was being built with Russian technology and equipment. At the time, post-Soviet Russia faced economic hardship and was eager to export major equipment such as nuclear technology. It struck a barter arrangement with China National Nuclear Corporation: China would not pay cash for the equipment; Russia would provide a $1.4 billion supplier credit, which China would repay with textiles, light industrial goods, household appliances, and other exports. The terms were attractive to China. By the late 1990s, under Premier Zhu, China’s foreign exchange reserves had improved dramatically, and paying $1.4 billion in cash was no longer a concern.

After hearing Ma Fucai’s report, Premier Zhu decided to accept the Russian terms—paying the $1.4 billion in cash instead of barter—in exchange for Russian agreement to build the pipeline, and dispatched me to lead a delegation to Russia to follow through. I went to Russia, stayed at our embassy in Moscow, and held discreet talks with various Russian ministries. A female director-general from Russia’s Finance Ministry accompanied us. I visited the Ministry of Economic Development, the Finance Ministry, and other departments, but I sensed the Russians did not share Ma Fucai’s message: on whether Chinese cash payment for Tianwan would buy Russian agreement to the pipeline, they were noncommittal—evasive and equivocating. The visit yielded no progress, and the matter fizzled out. My guess was that Ma Fucai had gotten his information from Yukos—perhaps it had been Yukos’s idea.

“Anda Line” vs. “Anna Line”: Japan Stirs the Pot

Opposition to the Anda Line then grew louder within Russia. The main objection was that the route passed near the southern end of Lake Baikal, the world’s largest freshwater lake, said to hold one-seventh of Earth’s terrestrial freshwater and a critically important Russian nature reserve. Critics feared a pipeline accident could pollute Baikal. Another objection was that Russia should maximize its national interest: the Anda Line would only deliver oil to China, while routes oriented toward Japan, South Korea, and other Asian markets deserved consideration.

The “Anna Line” then emerged. It was reportedly proposed by Japan. The route would run from Angarsk along the Baikal-Amur Mainline railway and the Sino-Russian border region to Nakhodka, Russia’s Far Eastern port. Unlike the Anda Line, the Anna Line stayed entirely within Russian territory, allowing Russia to export oil to other East Asian countries via the Pacific port of Nakhodka.

Japan, also a major oil importer, sought import diversification to reduce its dependence on Middle Eastern crude and was uncomfortable with the idea of Russia building a pipeline only to Daqing. From late 2002, Japan began actively lobbying Russia for the Anna Line. In the first half of 2003, then–Prime Minister Junichiro Koizumi met Putin twice specifically to discuss energy cooperation. In June, Japanese Foreign Minister Yoriko Kawaguchi and former Prime Minister Yoshiro Mori visited the Russian Far East and pledged $7.5 billion in loans for Siberian oilfield development and pipeline construction—conditional on the Anna Line being built. Japan also hoped energy cooperation with Russia would expand Japanese influence in the Far East and weaken the China-Russia strategic partnership. Bluntly put, Tokyo wanted to torpedo the Anda Line.

Owing to Japanese involvement, Russia reopened internal debate over the pipeline’s route. China and Japan engaged in a quiet tug-of-war over which route the Russian Far East pipeline would take.

Putin Finally Decides on the “Taina Line”—But Suspense Remains

In 2003, China’s government changed, with Hu Jintao becoming President. In May 2003, while SARS raged in China, Hu made his first overseas trip as head of state, attending the 300th anniversary celebration of St. Petersburg and the third SCO summit. I accompanied him. During the visit, Hu personally worked on Putin. The SCO Moscow Declaration was issued, the China-Russia strategic partnership was further consolidated, and the visit gave strong impetus to Russia’s decision to build the pipeline.

After more than two years of debate, on December 31, 2004, Putin personally decided to build the East Siberia-Pacific Ocean (ESPO) pipeline—the “Taina Line.” It moved the original Anda route more than 400 kilometers north, well away from Lake Baikal, resolving the long-running domestic environmental dispute. The Taina Line begins at Taishet in Irkutsk Oblast, runs more than 400 km north of Lake Baikal, follows the Baikal-Amur Mainline, runs along the Sino-Russian border from Skovorodino, and ends at the Pacific port of Nakhodka.

The Taina Line was to be built in two phases. Phase one would lay the Taishet-Skovorodino segment with a designed annual throughput of 30 million tons, while large oil storage facilities were to be built at Nakhodka. Skovorodino sits across the Heilongjiang River from Mohe in China’s Heilongjiang Province. The pipeline would cross the river at Tenda on the Russian side, after which China would build a pipeline from Mohe to Daqing. Phase two would lay the Skovorodino-Nakhodka segment with annual capacity of 50 million tons and expand the Taishet-Skovorodino capacity to 80 million tons.

The Russian federal government issued Order No. 1731, and on April 26, 2005, the Russian Ministry of Industry and Energy issued Order No. 91 approving construction of the Taina Line.

The decision came against a backdrop of strengthened mutual trust within the China-Russia strategic partnership. Russia recognized China’s strong desire for the pipeline; indeed, the pipeline had become a test of the strategic partnership and bilateral trust. It also reflected Russia’s domestic politics, economics, and social dynamics, its energy diplomacy and pursuit of national interest, and its broader posture on exporting oil to East Asia. But Russia had not yet committed to a branch line into China to Daqing. Some Russians even said: “If China wants oil, it can just import it from Nakhodka!” Building the pipeline was only the first step of a long march; tough negotiations and bargaining still lay ahead.

President Hu Jintao Points to Russia’s “Internal Cause” for the Stalled Talks

Looking back, Russia’s prolonged indecision had much to do with CNPC’s initial choice of Yukos. Khodorkovsky had been a Komsomol secretary who had “jumped into the sea” of business during privatization. He was politically at odds with the authorities and had political ambitions. In 2004, the Russian government began investigating Yukos for tax evasion, and Khodorkovsky was imprisoned.

When Hu Jintao visited Russia, he summoned Ma Fucai and me to his presidential suite to report on the pipeline negotiations. Hu and his wife Liu Yongqing were having dinner, and he gave me a piece of roasted sweet potato—I held it in my hand but didn’t dare eat. Ma emphasized that Japan’s interference was the main reason talks were stalled. I remember Hu’s response: “Don’t only look for causes externally—look for causes inside Russia too.” We didn’t fully grasp this at the time. The next day, Ma and Khodorkovsky still held a joint press conference, but only a deputy foreign minister attended on behalf of the Russian government—a clear signal of official coolness.

Ma Kai马凯 and You Quan尤权 Push Through the $6 Billion “Loans-for-Oil” Plan, Securing China’s Priority for the Branch Line

After the tax-evasion arrest of Khodorkovsky, Russia publicly auctioned 76.79% of Yugansk Oil & Gas, a Yukos subsidiary, on December 19, 2004, to settle Yukos’s tax claims. Baikal Finance Group won with a bid equivalent to roughly $9.3 billion in rubles. Rosneft then acquired all of Baikal Finance Group, becoming owner of the 76.79% stake in Yugansk. Yukos’s American shareholders sued the Russian government in the United States for expropriation. To raise funds, Rosneft proposed a “loans-for-oil” arrangement to China: it would borrow $6 billion, repaying through oil sales.

No Chinese bank had ever made such a large one-time loan to a foreign enterprise. Moreover, after the Soviet collapse, China’s public sentiment was generally pessimistic about the Russian economy and skeptical about Russian creditworthiness. A one-time $6 billion loan to Rosneft was hard to support in China’s financial circles.

Then-NDRC Chairman Ma Kai convened a coordination meeting. Only Ma Kai and I clearly favored extending the loan. We argued that China needed Russian oil, that imports from Russia were rising annually, and that if the loan was linked to Chinese oil purchases, the risk would be manageable. This was a chance to expand energy cooperation with Russia and break the pipeline deadlock. The China Development Bank, then headed by Chen Yuan, had long held to lending principles aligned with national industrial policy. Chen advanced a “development finance” approach, and CDB indicated it was willing to be the lead bank for the loan if the State Council so decided.

The NDRC reported the coordination outcome and our recommendation to the State Council, where it received leadership support. Then-Deputy Secretary-General of the State Council You Quan played an active coordinating role, persuading various departments to agree to extend $6 billion to Rosneft, linked to Chinese oil purchases. Commercial negotiations were left to CNPC, CDB, and their Russian counterparts.

On January 8, 2005, CNPC and Rosneft signed a long-term trade contract for importing 48.4 million tons of Russian crude. Under it, Russia would supply 48.4 million tons of oil by rail to China between 2005 and 2010, receiving the $6 billion loan from China and repaying it from oil sales proceeds. CDB agreed on commercial terms—the LIBOR rate plus 300 basis points, a not-low rate following fully commercial principles, with a six-year term. A dedicated bank account was established: Chinese payments for Russian oil went into the account, and CDB was repaid principal and interest from it on schedule. CNPC’s purchase price followed international benchmarks—Brent, WTI, Dubai—via a formula. During implementation, as LIBOR rose, adding 300 bps exceeded then-prevailing commercial loan rates, and at Russia’s request CDB reduced the spread to around LIBOR plus 200-plus basis points, roughly in line with commercial loans. Media speculation about the oil prices and loan rates was simply inaccurate. By 2011, Russia had repaid the entire $6 billion principal and interest, and China had bought 48.4 million tons of oil—a true win-win. CDB used this loan as a launchpad for its international finance business, which later became a major line.

Subsequently, Rosneft sought to go public and looked for strategic partners. CNPC and Sinopec had done the same when they listed—BP and Shell became strategic investors holding 10% stakes, only to sell when share prices rose. They reportedly profited roughly the equivalent of all their prior investment in China. CNPC and Sinopec were furious, calling this “unfriendly,” but the response was that this was normal commercial behavior. Rosneft turned first to CNPC for strategic investment. Ma Kai and I argued China should seize the chance to acquire a larger stake in Rosneft—something we had long pursued. But CNPC took a more commercially conservative view, judging the share price too high. Later, CNPC Vice President Zhou Jiping reported to Ma Kai and me that they had bought $500 million of Rosneft shares as a strategic investor, and that Rosneft was grateful. I was displeased and told CNPC right in front of Ma Kai: India bought $1 billion—we ought to be Rosneft’s strategic partner, and we had long wanted upstream Russian oil assets, so wasn’t this the moment? Why did we buy less than India? CNPC’s explanation was that the share price was overvalued. But after Rosneft’s IPO, with oil prices high, the stock rose. As the saying goes, “man proposes, but heaven disposes.”人算不如天算

There was also a side episode: oil and gas cooperation with Russia had been handled solely by CNPC, as we worried that if Sinopec or others joined, internal Chinese competition would be exploited by the Russians, to our disadvantage. The government, therefore, endorsed CNPC’s sole role and restrained the others. But Sinopec, also one of the “three barrels of oil,” was unhappy to be sidelined and quietly bought $200 million of Rosneft shares. Sinopec also secretly negotiated with Rosneft on another pipeline route from Russia through Mongolia to Yanshan near Beijing. This conflicted with our original goal of using Russian imports to offset Daqing’s decline, and risked derailing the Daqing pipeline. Then-Vice Premier Zeng Peiyan informed Sinopec Chairman Li Yizhong that Sinopec should stop negotiating the Mongolia route, and Sinopec withdrew. Still, Sinopec imported some Russian crude by rail through Mongolia to Yanshan Petrochemical. That was a minor side note.

“Loans-for-oil” warmed Russia’s attitude toward a branch line to China. On July 8, 2005, Putin said at a press conference in Scotland for the first time that Russia would prioritize a branch line to China when building its Far East pipeline. On September 7, Russian media reported that Putin had told Western journalists at the Kremlin: phase one of the East Siberia-Pacific pipeline would extend to Daqing, Russian oil would first flow to Daqing, and the Daqing branch line would come first—although the pipeline would eventually reach Nakhodka. The “China-first” principle was thus established.

On March 21, 2006, Putin visited China. Energy cooperation was a central issue, and the two sides signed a series of major documents, moving cooperation into a new, more substantive phase. Among the documents was a Minutes of Meeting between CNPC and Transneft, whose core content was that the Russian side would complete the project proposal and investment justification for the Skovorodino-China border segment and submit them to Russian authorities for review. After this, the companies worked intensively on surveying that segment, designing the Heilongjiang River crossing, and more.

Russia’s attitude warmed but remained ambiguous. Although Putin repeatedly mentioned a Chinese branch line, the Russian government had never formally committed to it in any official bilateral document. The pipeline’s fate remained uncertain.

Grueling Marathon Talks—Russia Keeps Adding Conditions

Negotiations with Russia were extremely difficult. The intergovernmental agreement was led on our side by me and on the Russian side by Deputy Energy Minister Yanovsky. Yanovsky was an expert-style leader—meticulous, careful, and patient. I greatly admired his professionalism, but “in war, each fights for his own master.” At the negotiating table, he scrutinized every word, haggled over every detail; with language barriers added, it was common to spend several hours on a single clause.

Russia knew well how much we wanted the pipeline and sought to maximize national interest by linking the pipeline to other items of Russian interest. For example, they wanted the agreement text to specify that Units 3 and 4 of the Tianwan Nuclear Plant would still use Russian Rosatom technology and equipment; they insisted not just on exporting crude but on Russian investment in a refinery and gas stations in Tianjin; they wanted to write in 15 million tons of Chinese coal imports from Russia per year; they wanted China to expand imports of Russian electricity; and so on. Because of long-standing mutual wariness, both sides were extremely cautious about wording: which sentence came first, which clause came first—endless disputes. Besides Chinese and Russian, the text required an English or French version as a counterpart. There were also disagreements over the arbitration court and governing law. So Yanovsky and I sometimes negotiated for a full day and night—dozens of hours straight.

In 2007, Hu Jintao visited Russia again to attend the “Year of China” in Russia. In meetings with Putin, energy cooperation was unavoidable. So CNPC representatives and I went to Moscow in advance to prepare, hoping the intergovernmental agreement could be signed during the visit. Yanovsky and I had another all-night marathon. I took a sleeping pill and was about to rest when President Hu arrived at the presidential hotel and immediately asked his secretary Chen Shiju to hear our update. The next morning I woke to find someone sleeping on the sofa in my room. I had a single room—who was there? It turned out to be my secretary, Fu Chaoqi. He told me I had fallen asleep at Chen Shiju’s meeting the night before, and he and then-NDRC Director-General of External Affairs Ma Xin had carried me back. Such was the toll of the negotiations.

After Putin Became Prime Minister, Pipeline Cooperation Was Unaffected

Work continued on every level. Transneft completed its investment justification for the Skovorodino-border segment, and on April 26, 2007, Russia’s General Administration of Construction Review approved it. On that basis, in June 2007 CNPC and Transneft signed a Memorandum on Accepting the Engineering Design of the Skovorodino-Border Segment. At the 9th China-Russia Energy Cooperation Committee meeting in July 2007, Russian Energy Minister Shmatko and I jointly hoped the companies would negotiate the intergovernmental pipeline agreement and submit proposals to both governments soon. China then provided a draft text. On March 2, 2008, Medvedev was elected President, with Putin as Prime Minister. After the Russian election and government reshuffle, in July 2008 Transneft completed engineering design for the Skovorodino-border segment and submitted it for Russian government approval. Preparations advanced steadily.

On a parallel track, talks on a new long-term Sino-Russian crude trade contract for the post-2010 period were also under way, replacing rail transport with pipeline delivery. The sticking point was price. Russian negotiators said reaching agreement on long-term contract pricing was a precondition for Russia to commit to the pipeline. Through difficult negotiations, the two sides gradually converged on the volume (10–15 million tons per year), start date (January 1, 2011), and term (10–20 years).

On August 18, 2008, I chaired a meeting to discuss progress on oil and gas cooperation. The companies still hadn’t agreed on pricing principles and formulas, but they agreed to complete contract talks soon, aiming to sign the long-term oil trade contract by the end of October. But on September 22, a meeting of senior CNPC and Rosneft leaders still failed to reach an agreement. The dispute centered on Russia’s insistence on pricing exports to China based on the Nakhodka price, hoping Nakhodka would become a new benchmark alongside Brent, WTI, and Dubai. China demanded that the export price equal Nakhodka minus the Skovorodino-Nakhodka pipeline transport cost.

Vice-Premier-Level Energy Negotiation Mechanism Established, with Wang Qishan as the Chinese Chair

At the state level, the two countries’ leaders attached great importance to energy cooperation. In May 2008, the two heads of state proposed a vice-premier-level energy negotiation mechanism. On July 26, then-Vice Premier Wang Qishan and Russian Deputy Prime Minister Sechin launched the mechanism in Beijing and held its first meeting. The mechanism operated on a non-fixed schedule, convened by either side as required for cooperation.

In late October 2008, Premier Wen Jiabao visited Russia and held the 13th regular Sino-Russian heads-of-government meeting with Prime Minister Putin. Such high-level meetings were standard occasions for advancing energy cooperation. To prepare, the Sub-Committee on Energy Cooperation (which I co-chaired with Shmatko) and the newly established vice-premier mechanism met successively in Moscow.

On the morning of October 23, 2008, Shmatko and I co-chaired the 10th meeting of the Sub-Committee. The focus was the pipeline. In drafting the meeting minutes, China hoped to clarify the pipeline at the governmental level and proposed signing an intergovernmental document during the upcoming premiers’ meeting—but Russia agreed only to let the companies continue engineering design and government approval.

Settling the Oil Pricing Formula

In a parallel venue, CNPC and Rosneft finally agreed on the long-term oil pricing mechanism, resolving the key obstacle. The export price would be the Nakhodka price minus Skovorodino-Nakhodka transport. The formula: P = N − T. The value of T remained undetermined. China had asked for a $10/barrel discount—clearly unrealistic, a typical opening gambit; even CNPC knew Russia wouldn’t accept it. Later Shmatko pressed me to set T at zero, which I, of course, could not accept. I responded: if T equals zero, why does the original formula include the “minus T” component? The matter remained pending. Later, through Wang Qishan and Sechin’s vice-premier talks, T was gradually narrowed and eventually agreed.

At the Initialing Ceremony, Negotiators Embraced and Wept

But the intergovernmental agreement remained in stalemate; Russia’s stance remained inscrutable. They seemed to be waiting for something.

On October 24, 2008, Russia scheduled no negotiations. Deputy Minister Yanovsky said he had nothing scheduled and asked whether I’d like to visit Rosneft headquarters. I said yes. At Rosneft headquarters, my hosts told me I had come at just the right moment: Deputy Prime Minister Sechin, Energy Minister Shmatko, and the heads of Rosneft, Transneft, Gazprom, and other top energy companies happened to be there discussing work. Would I like to meet them? I would. Through this “chance encounter,” I met Sechin at Rosneft headquarters.

Sechin was very warm. In the reception room, he told me: China needs oil, Russia has oil, and Russia is willing to export to China. But since he had assumed responsibility for energy as deputy prime minister, he had carefully reviewed the plan to export oil to China and concluded that Russia’s existing production capacity could not guarantee 15 million tons per year for 20 years. New oilfields had to be developed and new pipelines built. He had a map brought in and laid it on the floor, showing me Russia’s existing fields and proven reserves. He pointed to a spot near Taishet and said a new field there must be developed, and that the pipeline would have to cross swamps—construction would be difficult, requiring huge investment, and Russia lacked the capital. He proposed that if China would provide a $15 billion loan, Russia could build the pipeline at 15 million tons per year. The 2007 US subprime crisis had triggered the global financial crisis; oil prices had collapsed, the ruble had depreciated, and Russian companies indeed faced funding shortages. I said I would have to report this to Beijing. Back at the embassy, I reported home immediately. China convened emergency consultations, with Deputy Secretary-General You Quan coordinating views that very day.

Our delegation was housed in a diplomatic apartment near Moscow, where we tended a fireplace ourselves. Outwardly I admired the autumn scenery along the Volga, but inwardly I was anxious—I knew the domestic process, where multiple agencies often disagreed and delays could be costly. Much depended on the coordinator’s skill and the leadership’s resolve.

To my relief, around 3 a.m. Moscow time, word came back from Beijing: the loan was approved, conditional on the pipeline being finalized. My heart settled. I was grateful for the swift and decisive coordination at home. Unable to sleep, I went out alone to walk along the Volga and watched the sunrise turn the river crimson. After daybreak, I asked Embassy Counselor Pei Jiansheng to notify the Russian side; the initialing ceremony was held immediately. Veterans of the pipeline talks could not contain their emotions—an old Russian Energy Ministry hand lifted up our young Chinese interpreter, Cao Wei, and the two embraced and wept.

Russia Raises the Loan to $25 Billion

On October 26, 2008, accompanied by Deputy Secretary-General Bi Jingquan, Wang Qishan arrived in Moscow to meet Sechin under the vice-premier energy mechanism.

I made an embarrassing slip-up going to meet Wang Qishan at the airport. Counselor Pei Jiansheng arranged the vehicles and departure time. Worried about Moscow traffic, I had asked to leave early; Pei thought we had ample time. We got stuck on the road, unable to move forward or back—there was no real expressway from central Moscow to the airport. Wang’s plane had landed; I was still on the road. I improvised: jumped out, climbed a nearby pedestrian overpass, hailed a cab going the opposite direction, and rushed back to the hotel—arriving at the entrance just in time to welcome him.

The next day, in Wang Qishan’s meeting with Sechin, Russia raised the loan amount to $25 billion. After internal consultation, China agreed and asked the two energy ministries to implement the details.

Round-the-Clock Combat; The Bank’s Comments Made Me Furious

On the afternoon of October 27, 2008, I sat down with Yanovsky to negotiate the intergovernmental documents on the pipeline and the loan. Talks were held at the presidential hotel in a friendly atmosphere, but on textual details both sides argued meticulously.

By 4 a.m. on the 28th the text was basically agreed, and both lead negotiators could finally relax. Yanovsky invited me down to the lobby for a drink while the staff finalized printing. At 4 a.m. the lobby bar had no attendants. Yanovsky lit a cigarette; I, who never smoke, cheerfully took one too—it was a thin cigarette, the sort women might smoke—and we joked about then–US Energy Secretary Bodman. Yanovsky, in good spirits, suggested adding one more line to the text: that the two sides could also explore the possibility of Russia exporting 30 million tons per year to China. I was delighted to agree—I thought it was a bonus. But that line caused trouble.

After everything was settled, Yanovsky went home to sleep. At 6 a.m. Deputy Secretary-General Bi Jingquan毕井泉 obtained the draft Yanovsky and I had finalized to circulate within the delegation. A banking official thought the added final line was inappropriate. He argued that Russia’s commitment to export 15 million tons per year was tied to China’s $25 billion loan. By adding a possibility of 30 million tons, didn’t Russia implicitly want another $25 billion loan later? So the bank demanded the line be removed. But by then I couldn’t find Yanovsky—he had gone home—and I had to ask Counselor Pei to track him down and remove the line. I was furious. I bluntly told the bankers: through which department did you discuss buying all those US bonds? ” Moreover, the text said nothing about another $25 billion loan.

Last-Minute Reversal at the Great Hall of the People; Premier Wen Jiabao’s Meeting Delayed Three Hours

On October 28, 2008, Premier Wen Jiabao met Putin in Moscow for the 13th regular Sino-Russian heads-of-government meeting. That day, Shmatko and I signed, on behalf of our governments, the Memorandum of Understanding on Cooperation in the Oil Sector; CNPC and Transneft signed the Principle Agreement on the Construction and Operation of the Skovorodino-China Border Crude Oil Pipeline. During this visit, China and Russia essentially decided to build the pipeline—only one step remained before legally binding agreements.

Over the next nearly four months, governments, companies, and financial institutions held many more rounds of arduous talks on contract details and interest rates. On February 17, 2009, the third meeting under the vice-premier mechanism was scheduled; relevant documents were to be signed there, so on the eve of the meeting the companies and banks negotiated through the night to settle the texts.

On the morning of February 17, 2009, in the Great Hall of the People, Wang Qishan and Sechin led their teams in a meeting. The morning went well; all seemed smooth. Wang hosted a lunch for Sechin, with the signing planned for after the meal and Premier Wen to receive Sechin’s delegation at Zhongnanhai after the signing. But during lunch, Transneft President Tokarev suddenly reported to Shmatko that several points in the contract text could not be reconciled with the Chinese side. The key issue was that Rosneft and Transneft demanded separate loans and separate repayments—each with its own document, refusing to be combined into one. If not changed, the contract could not be signed.

Russia’s habit of last-minute changes was not new to me—I had to wonder if it was a negotiating tactic. Later, at Blagoveshchensk, when we negotiated the Tianjin refinery project, the same thing happened: everything was settled, the Russians arranged a riverboat dinner on the Heilongjiang, and we were to sign upon disembarking—but on the boat, Russia suddenly said it could not guarantee crude supplies for the Tianjin refinery and demanded the text be altered. I was puzzled: wasn’t selling crude to us Russia’s own goal? Otherwise, what was the point of a joint refinery in Tianjin? Over that one clause, we negotiated past midnight. Other unrelated company representatives grew sleepy and impatient; the Russians, taking the lead, started a “gambling game”—betting on whether we’d sign that night and at what time. Then-Shenhua Chairman Zhang Xiwu was at the meeting and put down a ruble for fun. Talks with RUSAL over a joint venture at Vanino on the Pacific went the same way: a JV text agreed in the morning, and at the afternoon ceremony Russia abruptly demanded reducing the Chinese share. I angrily walked out.

When I reported the latest snag to Wang Qishan, he said: “If you can’t agree, keep talking. We’ll wait.” The two energy ministers and the CNPC and Transneft heads then left the table to confer.

Sechin and the Russian ambassador pulled me aside, repeatedly urging China to accept splitting the deal into two documents. That itself was not a big issue, but convincing CNPC and CDB took time—CDB had to worry about how Transneft would secure repayment and would need new clauses. We waited three hours. I was on tenterhooks, anxious that the leadership was growing impatient. Because no one had expected new negotiations, the unusual scene emerged of company leaders and staff negotiating standing up. Sechin, the Russian ambassador, and I, plus one interpreter, talked standing beside the elevator. Once the differences were resolved and we waited for the companies to prepare the texts, Sechin’s usual stern look gave way to smiles and ease, and he even posed for photos with Chinese staff. Finally, after 5 p.m.—more than three hours later than scheduled—a grand signing ceremony was held at the Great Hall of the People. Shmatko and I, on behalf of our governments, signed the intergovernmental agreement. At the same time, the package was signed: the $25 billion financing loan, the pipeline construction and operation contract, and the long-term oil trade contract (under which, starting January 1, 2011, Russia would supply China 15 million tons of crude annually via pipeline for 20 years).

With Wang Qishan and Sechin as witnesses, 15 years of negotiation were finally sealed. Premier Wen’s meeting at Zhongnanhai was also delayed by three hours; the leaders’ immense patience deeply impressed me. The two countries’ leaders had made a political decision from the heights of long-term strategy.

That day, moved by emotion, I wrote a Pusaman lyric poem to commemorate the grueling negotiations.

菩萨蛮·庆中俄原油管道签约Celebrating the Signing of the China–Russia Crude Oil Pipeline Agreement
十五载跌宕博弈，Fifteen years of seesaw bargaining,
一昼夜鏖战斗智，One full day and night of hard-fought wits,
安大多诡谲，The Angarsk–Daqing line so full of twists,
泰纳藏玄机，The Taishet–Nakhodka route concealing hidden scripts.
政经相交织，Politics and economics intertwined,
外交暗角力，Covert diplomatic wrestling behind,
首脑亲运筹，Heads of state themselves the strategy designed,
远东布新局。In the Far East a new chessboard is now aligned.

The intergovernmental agreement on oil cooperation, the loan agreements, pipeline construction agreement, and oil trade contract took formal effect on April 21, 2009.

In June 2009, President Hu Jintao made another state visit to Russia. On June 17, meeting with Medvedev, he again emphasized: “Fully implement the intergovernmental agreement on oil cooperation between China and Russia, lay a solid foundation for long-term, comprehensive, and stable energy cooperation, actively advance cooperation in renewable energy, new energy, and other fields, and forge a comprehensive, integrated energy cooperation framework.”

After the oil pipeline breakthrough, mutual trust strengthened, and cooperation quickly expanded into natural gas, coal, electricity, and renewables. After the talks, Minister Shmatko and I signed MOUs on China-Russia gas and coal cooperation.

China and Russia Sign Cooperation Documents

In 2009, then–Vice President Xi Jinping led a delegation to Russia’s Far East, Moscow, and St. Petersburg, accompanied by me and Chen Yuan. During the visit, Xi continued to engage Russian leaders to develop and consolidate energy cooperation. Under the witness of Xi Jinping and Liu Yandong, the energy and regional cooperation documents were signed.

The pipeline was settled, but international political and economic gamesmanship remained complex. While deciding to build the pipeline to Daqing, Russia also moved to placate Japan, deciding to build a 7-million-ton-per-year LNG plant on Sakhalin as a joint venture with Japan to supply LNG. The Japanese Prime Minister personally attended the groundbreaking ceremony on Sakhalin, and Japan got its psychological compensation.

I Jokingly Scolded Counselor Pei Jiansheng for “Serving Two Masters”

After the signing, the construction agreements—defining obligations and responsibilities—still saw friction.

The lead Chinese negotiator was Jiang Qi蒋奇 of CNPC, a former economic counselor at our Moscow embassy. In Moscow, Counselor Pei Jiansheng裴建胜 handled day-to-day talks.

It was agreed that China would carry out the Heilongjiang River crossing, requiring our crews to cross to the Russian side frequently. Treating every crossing as a formal border exit would be too cumbersome. We asked Russia to designate a construction zone on its bank where our workers would operate as if domestically, without immigration formalities—effectively, Russian border control and customs would retreat behind the zone. Russia agreed. Frankly, were the situation reversed in China, this might not have been possible—it touches on territorial sovereignty. Next: Chinese construction equipment moved into that zone should not count as exports to Russia and should not be subject to import duties. Russia agreed (with the exception of tobacco, alcohol, and private cars). We also requested that Chinese personnel working in the zone be exempt from Russian taxes. Russia agreed. Frankly, these were lawyers’ inventions; once the first issue was solved, the rest followed, but the lawyers’ “rigorous” follow-up questions kept complicating things. Finally, the lawyers also demanded that any foreigners hired by the Chinese side also be exempt from personal income tax. Russia balked.

Talks stalled, and the issue came to me. Counselor Pei kept me informed throughout, but because he was seconded to NDRC from CNPC (Pei was fluent in Russian, a Tsinghua undergraduate, and held a PhD from a Moscow management university), his personnel file remained with CNPC, and he reported to CNPC first and largely conveyed CNPC’s position to the Russians. When the foreign-employee tax exemption came up, I lost my temper. I asked: where do “foreigners” come into this? He said: what if we hire foreign supervisors or engineers later? I said: then it should depend on whether that foreign employee’s country of nationality has a double-taxation agreement with Russia. We’ve spent 15 years on this pipeline; we won’t be held up by hypotheticals that may never occur. I joked-scolded Pei: don’t serve two masters. You’re now NDRC’s representative on this matter, so listen to me. If we do hire foreigners later, we’ll follow the tax treaty between their country and Russia.

That settled it.

“Looking Back, the Bleak Road; The Tide of the Heilongjiang Surges to Launch a Grand Vision”

On April 27, 2009, construction began on the Russian-side segment. On May 18, the Chinese-side segment began at Xing’an Town, Mohe. Deputy Secretary-General Bi Jingquan, I, and other ministry and CNPC representatives accompanied Wang Qishan to the Mohe groundbreaking ceremony. After landing at the airport, we took a train segment, then a long car ride, crossing a mountain pass. The season was as the song “Spring in the Northern Country” describes: birches were just budding, but as we crossed the pass, heavy snow fell; once past, the sky cleared, sunlight blazed, and as we drove toward the Heilongjiang riverside site, wind and clouds danced over the vast forest, with rays breaking through the clouds. Our worry about rain dissolved.

Su Shi’s Dingfengbo contains the line: “Looking back at the bleak place I came from, going home, neither wind and rain nor sunshine.” Recalling the 15 years of negotiations—like a road of storms and bleakness—we had finally reached a sunny groundbreaking day on the Heilongjiang. China-Russia energy cooperation was opening a new chapter. Unable to contain my excitement, I composed a Dielianhua poem.

Putin Personally Drives a Lada to the Opening Ceremony and Turns the Valve

In late September 2010, the pipeline was completed and entered trial operation. Russia held a grand opening ceremony at Skovorodino, opposite Mohe. I was invited to lead the Chinese delegation; CNPC Vice President Wang Dongjin attended. We first arrived at Blagoveshchensk across from Heihe, where Shmatko and I co-chaired a Sub-Committee meeting; a Russian aircraft then took us to Skovorodino. Deputy Prime Minister Sechin had arrived earlier. He invited me to wait at a roadside, saying someone would arrive shortly. A Lada drove up—Putin himself was behind the wheel. He personally opened the valve, symbolizing Russian crude beginning to flow through the pipeline to China. At the grand ceremony, I spoke on behalf of the Chinese government to offer congratulations.

The Significance of the China-Russia Crude Oil Pipeline

As a strategic artery for China’s oil and gas imports from the northeast, the pipeline begins at the Skovorodino metering station on Russia’s ESPO line, crosses the Heilongjiang at Tenda, reaches Mohe in Heilongjiang Province, and passes through 13 counties, cities, and districts in Heilongjiang Province and Inner Mongolia, ending at the Mohe head station of the Mohe-Daqing pipeline. The Russian-side segment is about 63.4 km long, the Heilongjiang crossing is 1.5 km, and the Chinese segment from Mohe to Daqing is 965 km. Phase one’s designed throughput is 15 million tons per year.

In line with the agreement, the pipeline officially began operation on January 1, 2011. Operations have gone smoothly, and Sino-Russian oil trade has grown steadily. By May 2017, China’s imports of Russian crude via the pipeline had exceeded 100 million tons. According to customs data, Russia became China’s largest source of crude oil imports in 2015, supplying 42.43 million tons that year; in 2016, China imported 52.48 million tons from Russia—a record high. To reinforce friendship, we proposed awarding Friendship Medals to 10 Russians who had played key roles in the pipeline’s decision-making and construction: Deputy Prime Minister Sechin, Energy Minister Shmatko, Deputy Minister Yanovsky, the heads of Rosneft and Transneft, and others. A grand ceremony was held during Premier Wen’s visit to Russia. Russia later reciprocated by awarding Friendship Medals to 10 Chinese.

Gratifyingly, in March 2013 the two sides also agreed to increase supplies to 30 million tons per year. To this end, China began building the Mohe-Daqing parallel line (the Skovorodino-Mohe segment was designed for up to 30 million tons per year), to be completed and put into operation by the end of 2017. The pipeline would then deliver 30 million tons of Russian crude annually—a long-term, stable source of supply for China.

In 2016, China imported 380 million tons of crude. Only three onshore pipelines existed—Sino-Russian, Sino-Kazakh, and Sino-Burmese. The Burmese line still transships Middle Eastern oil arriving by sea, so the genuine onshore arteries are only the Russian and Kazakh lines. Pipeline imports account for only about 10% of all crude imports; the remainder must come by sea. The strategic importance of the China-Russia pipeline is therefore evident.

The pipeline was negotiated for 15 years, with politics and economics intertwined. What ultimately made it possible was the foresight of the leaders of both countries; several generations of leaders personally guided energy cooperation with Russia and established the China-Russia strategic partnership. Tens of thousands of officials at all levels, oil-industry cadres, technicians, and workers contributed their efforts. Looking back, I salute them all with deep respect.

The article is co-authored by Li Jia 李嘉, Dean of the School of Economics and Lee Kong Chian Professor of Economics at Singapore Management University, and Zhang Dandan 张丹丹, Vice Dean and Professor of Economics at the National School of Development, Peking University. Both are well-regarded labor economists, and Zhang in particular has done extensive empirical work on China’s labor market using large-scale recruitment data

.

Their research, based on roughly 1.63 million job postings scraped from Zhaopin (智联招聘), constructed an AI-LLM occupational exposure index showing that white-collar jobs are disproportionately concentrated in high-AI-exposure occupations. More importantly, the higher the occupational exposure risk, the higher the entry barriers employers are setting for junior-level applicants. In other words, AI is closing off the on-ramps for young workers at the very start of their careers. This also echoes Zhuo Xian’s concern about the “learning by doing” ladder being broken, but here it is grounded in microdata from China’s actual labor market.

Li and Zhang argue that AI-driven labor substitution generates an aggregate demand externality; each firm internalizes the cost savings from automation but does not internalize the consequence that displaced workers, with reduced incomes, consume less, which eventually feeds back to the economy as weaker demand. It’s treated not just as a job-loss problem, but also as a market failure.

Li and Zhang lay out four distinct national approaches:

• Singapore: “State-led coordinated transformation” — integrating industrial upgrading, Company Training Committees through NTUC, and the new SkillsFuture Jobseeker Support Scheme into a single transition framework.

• EU: Procedural constraints via the AI Act, requiring prior consultation with worker representatives before deploying high-risk AI systems — slowing substitution to buy time for adjustment.

• US: Federal retrenchment under the Trump administration, combined with fragmented state-level regulation (California’s SB951, NYC’s algorithmic bias audits) — focused more on the fairness of AI substitution than on its pace.

• China: An emerging “governance-within-development” path that emphasizes the creation effect of AI rather than just limiting the substitution effect.

At the policy proposal level, Li and Zhang sketch out a “one fund, two pillars, three supporting measures” framework. The “one fund” is a dedicated AI Employment Compensation Fund, financed through fiscal allocations, surpluses from the unemployment insurance system, employer contributions, and possible surcharges tied to data and computing power revenues. And that fund should be used for targeted interventions, such as retraining displaced workers and support groups like young and low-skilled workers.

The two pillars then divide the intervention timeline. In the short term, an AI-exposure-based employment risk monitoring and early warning system, with more flexible working hours and income support arrangements in high-risk industries. In the long run, they propose institutionalizing a lifelong learning system through portable “skills accounts” and “learning accounts” that move with workers across jobs, alongside formally incorporating AI employment governance into the Employment Promotion Law and future AI-related legislation.

Tech progress does not determine social outcomes on its own. Whether AI broadens shared prosperity or deepens social divisions depends on how the government mediates the relationship between technology and labor. This line of thinking also reflects the Chinese scholarly understanding of the relationship between the state and technological progress: on the one hand, rapid technological development is welcomed and actively promoted, viewed as a key engine for boosting productivity and national competitiveness; on the other hand, the government keep vigilance to the employment shocks and social risks that technology may bring, with a strong emphasis on publishing new policies as a safety net for technological change, preventing technological transformation from tipping over into social disintegration.

Thanks to Professor Zhang’s kind authorization, I can share her piece in English.

How to Build an “Interactive Compensation” Public Policy Mechanism in Response to AI’s Impact on Employment

By Li Jia and Zhang Dandan

The new wave of artificial intelligence (AI) technology, represented by large language models, is evolving from an auxiliary tool into a key factor of production at an unprecedented pace, rapidly reshaping the structure of the labor market. The AI-LLM Occupational Exposure Index, constructed by our research team based on approximately 1.63 million online job postings from Zhaopin, reveals that white-collar positions are more heavily concentrated in occupations with high AI exposure risk. At the same time, the higher the occupational exposure risk, the more pronounced the rise in entry barriers that employers impose on applicants for junior positions. AI is not only reshaping occupational structures, but also reinforcing labor market stratification at the very starting point of career development.

Even more noteworthy, this round of technological disruption is overlapping with trends such as population aging, rising youth employment pressure, and the expansion of flexible employment, presenting new governance challenges to the “employment-first” strategy. Against this backdrop, an unavoidable core question emerges: when the pace of AI substitution systematically outstrips the creation of new jobs and the reallocation of labor, are traditional employment policy tools still sufficient? And how are countries around the world responding to this wave of AI disruption?

Aggregate Demand Constraints in AI Substitution and the Logic of Government Intervention

Before discussing specific public policy options, it is necessary to first address why government intervention in the governance of AI-driven employment restructuring is necessary.

When deciding whether to use AI to replace workers, each firm only calculates its own returns—such as how much wage cost it saves and how much operational efficiency it gains—while ignoring the “social cost”: once replaced workers experience income declines, their consumption capacity weakens accordingly, and the resulting contraction in demand ultimately feeds back into the entire economic system in the form of fewer orders and declining sales.

In other words, the cost-saving benefits from corporate automation are typically captured exclusively by the firms themselves, while the demand decline triggered by employment contraction is collectively borne by society as a whole. This means that under a fragmented, competitive market structure, the automation process driven by rational decision-making at the individual firm level is unlikely to produce a socially optimal outcome. The more competitive and dispersed the market, the harder it is for any individual firm to “internalize” the impact of its own behavior on aggregate demand, and the stronger the “aggregate demand externality” caused by employment displacement becomes.

What makes this round of AI technological revolution unique is that its “breadth” and “speed” simultaneously and significantly exceed those of most historical technological transitions, making this aggregate demand externality logic especially important.

Historically, technological revolutions usually had relatively clear industry boundaries. Even when some jobs were displaced, workers generally still had relatively sufficient time and space to migrate to other industries or occupations. For example, industrial automation primarily affected manufacturing jobs, while the spread of computers mostly transformed office administrative workflows.

This round of AI technology represented by large language models, however, exhibits highly horizontal diffusion. A large number of cognitive jobs—including copywriting, customer service, junior programming, drafting legal documents, translation, and research report writing—are simultaneously entering the scope of AI substitutability. Technological disruption is no longer confined to a single industry, but is unfolding synchronously across industries and occupational tiers.

At the same time, the speed of technological diffusion far exceeds historical experience. It took electricity about 30 years to permeate the industrial system, personal computers about 20 years to become widespread, and the internet roughly 10 years to spread. By contrast, once a large language model goes online, it can be rapidly deployed and replicated globally via cloud computing platforms and digital infrastructure.

This means that when AI’s pace of labor substitution systematically outstrips the economic system’s pace of creating new jobs and workers’ pace of retraining and occupational transition, the problems brought by technological progress are no longer merely short-term frictional unemployment, but may evolve into persistent aggregate demand shortfalls and social welfare losses.

Precisely for this reason, the “externality” perspective provides theoretical space for public policy intervention. The use of AI is not in itself a “deviation” from corporate norms; on the contrary, it is often a rational choice under the pressure of market competition. The problem is that when all firms move in unison toward “reducing labor costs,” individual rationality can accumulate into systemic social welfare losses through a mechanism akin to the “prisoner’s dilemma.”

This kind of aggregate demand externality, which the market mechanism itself cannot easily repair, is precisely one of the areas where institutional intervention is most justified. In this sense, establishing new cooperative mechanisms among government, firms, and workers around AI transformation has implications not only for social stability, but also for correcting externalities and improving overall economic efficiency.

International Perspectives: Different Approaches in Singapore, the EU, and the United States

Facing the impact of AI on the labor market, different economies are forming sharply different response paths. The differences reflect not only the policy choices themselves, but also each country’s institutional structure, governmental capacity, and traditions of labor-management relations.

1. Singapore’s “Collaborative Governance” Approach

Entering 2026, Singapore is accelerating the formation of a governance framework that links AI industrial upgrading with employment stability. Its core logic is not simply to restrict technological substitution, but to actively promote AI diffusion and industrial upgrading while cushioning technological shocks, enhancing labor adaptability, and maintaining social stability through national coordination, skills training, and employment protection.

Overall, Singapore’s AI governance can be summarized as a “state-led collaborative transformation” model: industrial policy, labor policy, and social security policy are not isolated from one another, but are integrated into a single national transformation strategy.

First, at the national governance level, Singapore has elevated AI to the core of its national competitiveness and industrial upgrading strategy. In the February 2026 Budget, Singapore announced the establishment of a “National Artificial Intelligence Council,” chaired by Prime Minister Lawrence Wong. This is the first time Singapore has elevated AI governance to the level of cabinet coordination, attempting to integrate the dispersed regulatory and developmental functions of the Infocomm Media Development Authority, the Monetary Authority of Singapore, and the Personal Data Protection Commission, while advancing national-level AI action plans in priority areas such as advanced manufacturing, logistics, finance, and healthcare.

This arrangement reflects Singapore’s fundamental positioning of AI: AI is not merely a technological tool, but key infrastructure for raising national productivity, driving industrial upgrading, and reshaping international competitiveness. Therefore, the governance focus is primarily on “promoting adoption” and “facilitating transformation,” rather than restricting technological diffusion.

At the same time, Singapore does not treat AI solely as an industrial policy issue, but simultaneously incorporates it into the labor market governance framework.

In April 2026, the Ministry of Manpower, the National Trades Union Congress (NTUC), and the Singapore National Employers Federation (SNEF) jointly established the “Tripartite Employment Council” as a unified coordination point for the employment effects of AI transformation. This mechanism is built upon Singapore’s long-standing tradition of “tripartite consultation” among government, business, and unions, with the goal of minimizing the employment impact of technological substitution while promoting corporate digitalization and AI transformation.

An important foundation of this system is the “Company Training Committee” institution, which the NTUC has been advancing since 2019. To date, Singapore has established more than 3,800 Company Training Committees, covering more than 300,000 blue-collar and white-collar workers; among these, AI-related training programs doubled in 2025 compared with the previous year.

A key feature is that training is not “external re-education” detached from corporate needs, but is directly embedded in the firm’s internal transformation process. Firms, unions, and government jointly identify which positions will be affected by AI and which skills need upgrading, and they organize training and job-transition arrangements around actual production processes. In other words, Singapore seeks to push “corporate upgrading” and “worker transformation” forward in parallel, rather than waiting until jobs disappear before providing ex-post relief.

Therefore, the logic of Singapore’s AI governance is not “preserving jobs,” but “preserving employability.” This concept was reflected in Prime Minister Lawrence Wong’s emphasis at the May 1, 2026 Labor Day Rally that “We may not be able to save every job, but we must protect every worker.”

This approach is also reflected in institutional integration. The 2026 Budget announced that the former “Singapore Workforce Development Agency” and “SkillsFuture Singapore” would be merged into a unified “Workforce and Skills Agency,” creating—for the first time at the institutional level—a single entry point for both “job seeking” and “skills learning.” Against the backdrop of AI-accelerated occupational restructuring, employment services, vocational training, and skills upgrading can no longer be treated as separate policy modules, but must form a continuously connected “employment-training-reemployment” cycle.

At the same time, Singapore is also beginning to establish employment buffer and relief mechanisms under AI transformation.

The “SkillsFuture Jobseeker Support Scheme” was fully implemented in 2025–2026. This is the first time in Singapore’s history that a cash bridging support mechanism has been established for involuntarily unemployed workers. Eligible unemployed workers can receive up to S$6,000 in phased support over six months.

Notably, the scheme sets a threshold of “an average monthly income of no more than S$5,000 over the past year,” which roughly corresponds to the median income of full-time resident workers in Singapore. This means that the scheme mainly covers low- and middle-income workers, with its core objective being to reduce the short-term income risks faced by vulnerable groups during the process of technological transformation.

Thus, from industrial upgrading, corporate training, and worker skills transitions to employment relief, Singapore is in effect building a relatively complete AI transformation governance system: the government is responsible for overall coordination, firms are responsible for implementing transformation and training, unions are responsible for organizing workers and articulating their interests, and the fiscal system bears part of the transformation costs.

2. The EU: Slowing the Impact of Substitution Through Procedural Constraints

Unlike Singapore, which emphasizes the simultaneous promotion of technology adoption and labor transformation, the EU has chosen to incorporate AI into a high-risk governance framework through horizontal legislation and ex-ante regulation, raising the institutional cost for firms to engage in large-scale labor substitution through procedural constraints.

The core carrier of this approach is the “EU AI Act,” formally adopted in 2024. The Act will become fully mandatory in August 2026, and for the first time, it provides systematic regulation through unified legislation of the use of AI in the workplace.

The Act explicitly classifies AI applications involving labor relations—such as recruitment screening, job assignment, performance evaluation, promotion and dismissal decisions, and employee behavior monitoring—as “high-risk AI systems.” Before deploying such systems, firms must fulfill obligations such as ex-ante risk assessment, algorithmic bias testing, technical documentation retention, design of human oversight mechanisms, and ongoing compliance review.

Particularly important is the requirement that before officially using a high-risk AI system, firms must notify worker representatives and affected employees in advance.

This means that the EU has effectively established a “prior consultation mechanism” for AI deployment at the institutional level. AI is no longer merely an internal corporate technology upgrade decision, but has been incorporated into the labor relations governance framework. Even if firms possess the technical capacity and economic incentives, they must procedurally address concerns regarding worker rights, transparency, and fairness.

In a sense, the EU is not directly preventing AI substitution, but rather attempting to slow the pace of substitution by raising institutional friction and procedural costs, thereby buying time for labor market adjustment.

A similar approach is reflected in the “EU Platform Workers Directive,” which took effect at the end of 2024. The directive, for the first time at the level of EU law, stipulates that platform workers cannot be dismissed or subjected to adverse treatment solely on the basis of automated algorithmic decisions; decisions involving major labor rights must retain “human intervention” and appeal mechanisms. EU member states must complete domestic transposition by the end of 2026.

It can be seen that the EU’s institutional focus on AI and employment is not primarily on “promoting technological diffusion,” but rather on “procedural legitimacy in the use of technology.” Behind this lies the core idea of Europe’s long-standing “social market economy” tradition: technological efficiency cannot automatically take precedence over worker protection.

The limitations of the EU approach are equally evident. The EU is relatively active in “limiting the pace of substitution,” but its institutional capacity in “bearing the consequences of substitution” remains limited. The EU’s strengths are primarily reflected in ex-ante regulation, information disclosure, and procedural constraints, but its existing policy tools give insufficient consideration to the income losses, reemployment transitions, and aggregate demand shocks that AI substitution may cause.

In addition, the EU approach also faces the typical problem of “high consensus thresholds.” AI governance involves complex tradeoffs among member states’ industrial interests, labor regulations, and competitiveness in innovation, making it difficult for the EU to form stable and consistent policy consensus internally. Many seemingly ambitious institutional designs have repeatedly encountered member state disagreements and industry resistance during implementation.

A typical case is the “AI Liability Directive” proposed by the European Commission in 2022. The proposal sought to lower the burden of proof for AI victims and establish a unified liability framework for harm caused by AI systems. However, because member states and the business community could never reach agreement on the boundaries of liability, the proposal was formally withdrawn in 2025.

This reflects a deeper tension in EU AI governance: its institutional design tends to carry strong normative idealism, yet under the pressures of rapid technological evolution and global industrial competition, maintaining a balance between “regulation” and “innovation” remains a persistently difficult policy challenge.

3. The United States: Federal Retrenchment and Fragmented State-Level Governance

The United States exhibits a pattern of “light federal intervention and decentralized state-level advancement” on AI and employment issues.

After 2025, the policy priorities of the US federal government have shifted back toward “technological leadership,” “reducing regulatory burdens,” and “enhancing national competitiveness,” while issues of worker protection have clearly receded into a secondary position.

The federal government’s attitude on the issue of “AI and worker protection” has clearly contracted. In January 2025, the Trump administration formally revoked the Biden-era executive order that had designated “AI’s impact on the labor market” as a federal priority. That executive order had required relevant agencies to conduct assessments of AI-induced job losses, establish labor consultation mechanisms, and study the long-term effects of automation on employment and wage structures. Under the Trump administration’s new executive order framework, much of this content has been substantially scaled back.

The “America’s AI Action Plan” released in July of the same year proposed more than 90 federal-level AI policy actions, covering areas such as infrastructure, energy, computing power, defense, export controls, and research support. Content related to the labor market, by contrast, was only briefly mentioned as an ancillary issue, with no new fiscal budget or systematic institutional arrangements.

To some extent, this reflects the basic orientation of US AI policy: prioritizing technological innovation and industrial competitive advantage, rather than first establishing large-scale employment buffer mechanisms.

From the second half of 2025 to early 2026, multiple bipartisan bill proposals have emerged in the US Senate, including requiring large firms to disclose AI-driven layoffs, establishing a national-level AI workforce research center, and strengthening firms’ obligations to notify employees about automation deployment. However, as of May 2026, most of these proposals have not yet completed formal legislative procedures.

In the United States, what is genuinely driving concrete restrictive measures are certain state governments and local regulatory bodies.

California’s SB951, the “Worker Technological Displacement Act,” currently under deliberation, is somewhat representative. The bill would require firms conducting large-scale layoffs due to AI or automation to give workers at least 90 days’ notice in advance and submit a technological substitution statement.

New York City has moved even earlier. Since 2023, New York City has formally required firms using automated recruitment tools to conduct annual “algorithmic bias audits” and to disclose to job applicants the use of AI recruitment systems. Subsequently, Illinois, Colorado, and other jurisdictions have also advanced local legislation related to workplace AI.

Overall, however, most current US state-level regulation still focuses on “AI should not produce discriminatory outcomes”—such as algorithmic bias, fairness in recruitment, and information transparency—rather than truly addressing the deeper issue of “the scale of AI substitution itself.”

In other words, the US institutional framework is currently more concerned with “whether AI is fairly replacing labor” than with “whether AI is replacing labor too quickly.”

Behind this difference lies a long-standing feature of the US political-economic system: compared with Europe, the United States is more inclined to view technological progress as a matter of market competition and innovation efficiency, rather than primarily incorporating it into frameworks of social protection and redistribution. The federal government has always remained cautious about directly intervening in corporate automation decisions, while labor market adjustment relies more heavily on local governments, market mechanisms, and individual workers’ adaptive capacities.

Therefore, the US model may achieve higher efficiency in technological diffusion and innovation incentives, but the cost is that the employment risks and income volatility brought by AI shocks must, to a greater extent, be borne by individual workers themselves.

4. China’s Policy Direction: From “Shock Response” to “Interactive Compensation”

China’s understanding of the relationship between AI and employment is accelerating into an institutionalized stage.

In recent years, policy formulations on “AI and employment” have clearly shifted from “paying attention to impacts” to “active governance.” In May 2024, General Secretary Xi Jinping, during the 14th group study session of the Politburo, clearly stated the need to “strengthen the positive employment-creating effects of new technologies and prevent the substitution effect from being released in a concentrated way over the short term.” In the same year, the “State Council’s Opinions on Deeply Implementing the ‘AI Plus’ Initiative” proposed establishing a mechanism for assessing the employment impact of AI applications, and preventing and mitigating technological shocks. Subsequently, the Third Plenary Session of the 20th CPC Central Committee and the related arrangements in the “15th Five-Year Plan” further emphasized “promoting high-quality and full employment” and “building an employment-friendly development model,” and proposed a systematic response to the impact of new technological changes on employment structure.

Unlike some countries that mainly focus on “restricting substitution,” an important feature of China’s current policy logic is its greater emphasis on “enhancing the creation effect.”

In other words, the policy objective is not merely to slow AI’s impact on existing jobs, but more importantly to accelerate the formation of new employment demand and income sources through industrial expansion, the cultivation of new business forms, and skills upgrading—aiming to make technological progress a “blood-generating mechanism” rather than merely a “substitution mechanism.” This approach is highly consistent with China’s long-standing governance logic of advancing development and stability in tandem.

Against this backdrop, China’s policy framework around AI and employment is also gradually moving from fragmented response toward systematic construction. Its core direction is not simply to restrict technological diffusion, but to establish, while promoting AI development, a comprehensive governance system covering risk assessment, employment buffering, skills transitions, and social security adaptation, so that technological progress and labor market adjustment can be mutually coordinated.

From an international comparative perspective, China’s exploration is forming a path that places greater emphasis on “governance within development”: different from the US, which relies relatively heavily on spontaneous market adjustment; different from the EU, which focuses on procedural regulation; and different from Singapore, which relies heavily on tripartite consultation mechanisms—China seeks to maintain the pace of technological diffusion while simultaneously incorporating employment stability, skills upgrading, and social security into its national development framework.

Against this backdrop, the next step in policy design may need to shift further from “passively responding to technological shocks” to “actively building interactive compensation mechanisms.”

The core of so-called “interactive compensation” is not simply income redistribution, but the establishment, at the institutional level, of more stable adjustment mechanisms between the beneficiaries of technology and those harmed by technology, allowing the efficiency dividends created by AI to partially flow back to the worker groups bearing the costs of transformation, thereby easing the structural imbalances during technological diffusion.

Around this goal, a policy framework of “one fund, two pillars, three supporting measures” can gradually take shape.

First, “one fund” refers to exploring the establishment of a dedicated AI employment compensation fund. Its funding sources can adopt a diversified financing model, including fiscal support, surplus from unemployment insurance, corporate contributions, and supplementary revenue arrangements related to data and computing power. The focus of the fund is not on universal cash distribution, but on targeted support for job stabilization and expansion, retraining for job transitions, reemployment assistance, and support for key groups, so that part of the technological dividend can flow back to labor market adjustment.

Second, the “two pillars” correspond to short-term buffering and long-term institutional construction.

In the short term, an employment risk monitoring and early-warning system based on AI exposure should be established; more flexible working-hour and income support arrangements should be explored in high-risk industries; and more precisely targeted transition support should be provided for key groups such as young workers and low-skilled workers.

In the medium to long term, it is necessary to institutionalize lifelong learning systems and skills accumulation mechanisms. For example, arrangements such as “skills accounts” and “learning accounts” could be explored, allowing training resources to follow workers as they move; meanwhile, AI employment governance should gradually be incorporated into the framework of the “Employment Promotion Law” and future AI-related legislation, to enhance policy continuity and institutional stability.

Finally, the social security system must be made to adapt to the labor structure of the AI era. Priorities include improving protections for flexible employment and platform workers, expanding the role of unemployment insurance in supporting skills transitions and occupational shifts, and further exploring how to more stably channel the productivity dividends brought by AI back into the pension and social security systems.

Ultimately, AI’s impact on employment is not merely a technological upgrade, but rather an institutional test of “how growth is distributed.” Technological progress does not automatically determine social outcomes. What ultimately decides whether AI expands social prosperity or aggravates social divisions still depends on how institutions coordinate the relationships among technology, capital, and labor.

Authors: Li Jia is Dean of the School of Economics at Singapore Management University and Lee Kong Chian Chair Professor of Economics; Zhang Dandan is Vice Dean and Professor of Economics at the National School of Development, Peking University.

More to read:

The East is Read

Dandan Zhang: China’s factory workers go gig

Dandan Zhang is a Professor in Economics (with tenure) and Deputy Dean (in research, internal and international cooperation) at the National School of Development (NSD), and Deputy Dean of the Institute of South-South Cooperation and Development, Peking University…

Read more

a year ago · 23 likes · Yuxuan JIA and Siqi Lin

The East is Read

Dandan Zhang on why mandating fewer working hours in China may be premature

Few ideas sound more sensible than giving overworked Chinese workers more rest. Surveys suggest that non-agricultural employees in China were averaging more than 50 hours a week in 2023, while fieldwork indicates that gig factory workers put in 10 to 12 hours a day, six or seven days a week…

Read more

3 months ago · 11 likes · Zhu Yutao and Yuxuan JIA

The main outcomes include:

• A constructive consensus on tariff arrangements

• The establishment of a Trade Council and an Investment Council — the Trade Council will focus on tariffs on imported goods, while the specific mandate of the Investment Council has yet to be disclosed and may still be under negotiation

• A two-way breakthrough on agricultural tariffs and market access: the US side will move to address issues such as the automatic detention of Chinese dairy and aquatic products, exports of Chinese bonsai (with growing medium) to the US, and the recognition of Shandong as an avian-flu-free zone; meanwhile, China will push forward on US beef facility registrations and poultry imports

• Mutual reductions in agricultural tariffs

• China to purchase aircraft from the US, with the US committing to safeguard the supply of engine components for those purchases. (During last year’s trade war, Washington suspended the export licenses that had allowed GE to sell engines to China, which we needed to develop the C919.)

It's worth noting that the official communique uses the qualifier “preliminary outcomes,” signaling that negotiations are still ongoing. The text also explicitly states that “the two sides are still consulting on the details of the relevant outcomes,” which explains why no concrete numbers appear in China’s announcement. We'll likely have to wait for further information from both sides in the coming days. On a deeper level, releasing a preliminary outcomes list before the two sides have fully aligned on specifics also seems aimed at managing external expectations around Trump's China visit.

(Just before I publish this, the White House published its fact sheet-including an initial Chinese purchase of 200 American-made Boeing aircraft, the first such commitment since 2017, Chinese commitments to purchase at least $17 billion per year of U.S. agricultural products in 2026 (prorated), 2027, and 2028, on top of the soybean purchases agreed in October 2025, the renewal of expired listings for more than 400 U.S. beef facilities, the resumption of U.S. poultry imports from avian-flu-free states, and Chinese undertakings on rare earth and critical mineral supply chains covering yttrium, scandium, neodymium, and indium.) On political achievement, it put Iran in a high position. It’s not surprising he wants to get China on board, but whether it's about denuclearization or opening the Strait of Hormuz, and against the charge tolls. It's China's long-term position, so I don't think China would alter its Middle East stance.

Subscribe now

No surprise Chip is not included; it’s connected to the AI race, and neither side is there yet; risk management would be good enough for this time.

What exceeded my expectations was the creation of the Investment Council. In the working-level exchanges leading up to these talks, the US line had consistently been that it was “not yet ready” to engage China on bilateral investment issues. Yet an agreement on this point was reached in Busan, which I read as an indirect signal that the negotiations themselves went relatively smoothly, without major friction. In his May 14 CNBC interview, Bessent explained that the Investment Council’s core purpose is to pre-identify non-strategic, non-sensitive sectors suitable for Chinese investment, so that those investments wouldn’t need to be submitted to CFIUS for review.

The market didn't react well to Trump's announcement that China would buy 200 Boeing aircraft (China hasn’t confirmed this yet), and Boeing’s stock slid as a result. I never expected a so-called “Grand Bargain“ to begin with. After a year of trade-war escalations and de-escalations playing out on repeat, the two sides are still very much in “crisis management” mode, with low trust level, landing a mega-deal simply isn’t realistic.

My expectation going into these talks was that if the two sides could maintain stability, achieve a breakthrough on coordination mechanisms, and tack on a few trade deals in non-security-sensitive areas, that would already count as a success. Looking at the results, the two sides reached a consensus on framing the relationship as “constructive strategic stability” and established two new coordination mechanisms in trade and investment. So from China's perspective, this round of consultations is, without question, a positive development.

Below is the translation I made based on MoFCOM's release

MOFCOM Spokesperson Answers Press Questions on Preliminary Outcomes of China-U.S. Economic and Trade Consultations

Q: On May 14, the heads of state of China and the United States held a meeting and announced that the two countries’ economic and trade teams had achieved overall balanced and positive outcomes. Could the Ministry of Commerce share more details about the results of these consultations?

A: The heads of state of China and the United States held a meeting in Beijing, where they had in-depth discussions on economic and trade issues, charting the course and providing strategic guidance for the further development of bilateral economic and trade relations. President Xi Jinping noted that the essence of China-U.S. economic and trade relations is mutual benefit and win-win cooperation. The overall balanced and positive outcomes reached by the two economic and trade teams are good news for the people of both countries and for the world, and both sides should work together to safeguard the hard-won positive momentum.

On May 13, the Chinese and U.S. economic and trade teams held consultations in South Korea, laying the groundwork in the economic and trade arena for the meeting between the two heads of state. Guided by the important consensus of the two presidents and upholding the principles of mutual respect, peaceful coexistence, and win-win cooperation, the two sides engaged in candid, in-depth, and constructive exchanges on resolving each other’s economic and trade concerns and further expanding practical cooperation. Following this, the two sides held intensive consultations on the specific content of the outcomes and reached positive consensus.

The preliminary outcomes reached by the two sides in the economic and trade field mainly include the following:

First, the two sides will continue to implement the outcomes of earlier consultations and have formed positive consensus on relevant tariff arrangements.

Second, the two sides have agreed to establish a Trade Council and an Investment Council to discuss their respective concerns in the areas of trade and investment. Through the Trade Council, the two sides will discuss issues such as tariff reductions on relevant products, and have agreed in principle to reduce tariffs on products of concern to each side on a comparable scale.

Third, the two sides will resolve or substantially advance the resolution of certain non-tariff barriers and market access issues regarding agricultural products. The U.S. side will actively work to address China’s long-standing concerns regarding the automatic detention of dairy products and aquatic products, the export of medium-grown bonsai to the U.S., and the recognition of avian influenza-free zones in Shandong, among others. The Chinese side will likewise actively work to address U.S. concerns regarding the registration of beef facilities and the export of poultry meat from certain states to China.

Fourth, the two sides have agreed to promote the expansion of two-way trade in areas including agricultural products through arrangements such as mutual tariff reductions on a certain range of products.

Fifth, the two sides have reached arrangements concerning China’s procurement of aircraft from the U.S. and the U.S.’s assurance of the supply of aircraft engines and components to China, and have agreed to continue advancing cooperation in related areas.

The positive outcomes of the China-U.S. economic and trade consultations demonstrate that, by upholding the spirit of mutual respect, equality, and reciprocity, and through dialogue and cooperation, the two sides are able to find solutions to their problems. The two sides are still consulting on the specific details of the relevant outcomes. Following the direction of consensus set by the two heads of state, the economic and trade teams of both countries will work to finalize the outcomes as soon as possible and jointly ensure their implementation, so as to inject greater certainty and stability into future China-U.S. economic and trade cooperation and the world economy.

More to read:

Geopolitechs

What Did the Trump–Xi Summit Actually Achieve?

Overall Assessment of the Visit…

Read more

2 months ago · 9 likes · Geopolitechs

Understanding China

Xi-Trump Summit:a direction, not a destination

On the morning of May 14, the American delegation in the Great Hall of the People was waiting. Secretary of State Marco Rubio stood beneath vast ceiling of the reception hall, head tilted upward, studying its ornamental reliefs with an expression somewhere between admiration and restlessness. Nearby, Elon Mush held his phone at arm’s length, panning slo…

Read more

2 months ago · 2 likes · ClarenceGu