Welcome back to the AI Power Podcast. I'm Jack Conland filling in for Adam Goodwin. In today's episode, we'll focus on the upcoming summit between President Trump and Chairman Xi Jinping and the role that artificial intelligence will play. We'll first give an overview of how each side is doing on the AI front, what their chief concerns are, and how they're feeling heading into the meeting. Then we'll turn to the summit itself: what each side wants, where they're stuck, and what could realistically come out of the summit. And here as always to help us make sense of these developments is Greg Allen. Greg, how are you doing?
I'm doing great, Jack. It's going to be a heck of a week. There's a ton of stuff to unpack here. So hopefully everybody out here is looking for a China mega episode, because that's what you're about to get.
And let's get right into it. Before we get to the summit itself, though, I want to understand where each side stands going in. Let's start on the Chinese side with Beijing. Over the last few years, the US has tried to slow China's AI progress primarily with export controls, and yet Chinese AI keeps advancing. How do you think Xi is feeling heading into these talks?
So, I think Xi Jinping is probably feeling pretty good heading into these talks. And the basic reason is that he has things that he wants from President Trump, but he doesn't have things that he desperately needs from President Trump. And what I mean by that is, if you think about all the pressure facing the Chinese system — very high youth unemployment, slowed economic growth — those are not the metrics that Xi Jinping is grading himself most critically on. He's really focused on technological competitiveness and whether or not he's making progress in the key industries and advancing towards self-sufficiency in those industries. That's the same basket of priorities he's had since he came to power in 2012. And I think everything that he's seen happen in the global economic system and in global geopolitics since then has convinced him that he had it right the first time, and they're actually making important progress on those metrics.
So let's think specifically on AI. I think there's basically four ways whereby China has improved its AI posture and its competitiveness vis-à-vis the United States that overall make Xi Jinping feel like he has a pretty strong hand going into these negotiations. So the first is, the biggest threat that China faces in AI is those export controls cutting off access to compute and chip equipment. But what they have demonstrated is a strong ability to take advantage of both legal and illegal measures in order to continue gaining access to US compute. That's the first thing. The second thing they're doing is, even though they're still significantly behind, they're also making noteworthy progress on building their own chips and chip tools. The third thing is that Chinese AI models are still really appealing to not just folks in China, but folks around the world and other countries that matter to US geopolitical calculations. Fourth and finally is just that distillation is working as a rapid catch-up strategy, and the United States does not have a great rebuttal to it for all of its bluster so far. So I think those are the four things that are in the back of Xi Jinping's mind as he walks into the negotiations with President Trump.
Let's take those one at a time. Starting with the first one. You mentioned illegal and legal access to American chips that continues to fuel the Chinese AI push. How is China still getting access to these chips?
Well, I think we saw a headline just a couple weeks ago in the New York Times that came out on September 6th that reported that Inspur, a major Chinese data center company, one of the largest makers in the world of computer servers and therefore a key vector of accessing Nvidia compute from an integration perspective. They are on the entity list, but they are still able to buy massive amounts of American compute because of a ridiculous restriction that allowed them to continue buying even when they should not have been able to, at least from a strategic consideration. So, the key point here is that Inspur has a subsidiary called AIS that got $5.6 billion worth of advanced technology as a US subsidiary where it was able to get around its entity listings and then export that to Southeast Asia, and among all of those technology imports included $3 billion worth of Nvidia Blackwell chips.
So remember, the Trump administration said that their policy was: we're going to allow H200 sales, we're going to ban Blackwell sales, and in the future we're going to ban Vera Rubin sales. That was the Trump administration theory of the case. But you have to ask yourself, what is the gap here between the policy and the process? What do I mean by that? Let's just say hypothetically that we had a law that said that every American is entitled to the right to vote. Okay? And that is the policy: every American has the right to vote. But then we have a process. And the process is there's only one place to register to vote in all of the United States, and it's a single DMV in rural Alaska that is only accessible via private jet. Well, then we have a policy that says everybody's allowed to vote, but we have a process that says only people who have easy access to private jets are in practice going to be able to vote.
And so now take that analogy and apply it to export controls. We have a policy that Chinese companies cannot get Blackwell chips. But the process that we've had for enforcing that has been so pathetic in so many different ways. You have to ask yourself, is letting folks get away with smuggling — is that actually the secret policy? I mean, that's how bad it is here. Which again is not to say that the export controls have not had a strategic impact. I think they have, but it goes a long way towards explaining how we got here where China is still able to do it.
Now, keep in mind that Inspur example that we were just talking about — the Inspur example, them buying those chips from Nvidia — that was all legal. It only became illegal if it made its way into China, which is my assumption that it in fact did as part of the smuggling ring reported by the New York Times. So, that's one legal tool that China's been able to exploit very successfully. The second one is renting what it cannot buy, which is advanced computing capacity from cloud computing providers across Southeast Asia. And you know, like who is that actually helping? Who is this remote access to NVIDIA chips via cloud computing helping? Well, according to reporting by CNBC, quote, 'Chinese hyperscalers, including ByteDance, Alibaba, and Tencent, have reportedly accessed compute power from NVIDIA chips remotely via other Asian nations, including Thailand, Malaysia, and Japan. ByteDance was working with Singapore-headquartered Aolani, a cloud provider with Nvidia chips, to access compute in Malaysia,' according to a source familiar with the matter.
Aolani told CNBC it worked, quote, 'with a global and diversified customer base spanning customers from North America to Asia.' Then it goes on to say, 'The companies we service do not have ownership, potential future claim, or physical access to the chips that power our solutions. Any permitted access to our services, infrastructure, or technology is fully compliant with all applicable regulations,' the spokesman added. So that's a person basically saying — they didn't say China, but I think between the lines you can assume they mean China. And they're basically saying, like, look, it's legal. If you don't want us to do it, make it illegal. And that's unfortunately where we're at.
It's important to note that these companies are not small fish. You have Inspur, which is listed as an elite Nvidia partner. You have other firms who have been involved in this third-party access like Mega Speed, which was also investigated for chip diversion, but they still provide this entirely legal third-party access through countries like Malaysia and Indonesia that are building up massive compute capacities that China is renting out, again, in a completely legal sense.
Yeah. I mean, you put it well, right? Maybe folks who listen to this podcast have never heard of Inspur, but anybody in the data center industry around the world, I guarantee you, has heard of Inspur. And there's a sense in which a company like Inspur is almost too big to fail because they make up such a significant share of global rack integration.
They can effectively say like, hey, if you want to entity list us, if you want to try and destroy us, like that's actually going to have consequences for downstream compute availability and demand, not just in China, but around the world. So, if you do that, you're not just picking a fight with China, you're picking a fight with all of China's customers. This is the exact same problem that the United States faced when they went after Huawei — that these companies are so big that it's just not trivial to punch at them even if you have a good legal justification for doing so.
So, one final thing is just to like think about where is China domestically in computing capacity, and in that regard we actually got some interesting data from China's Ministry of Industry and Information Technology, MIIT. So the 15th five-year plan for the information and communication sector was just released on September 7th, and it set a target for 9,800 exaflops of intelligent computing capacity by 2030 and called for 3.8 trillion yuan, which is equivalent of about $530 billion US, in cumulative information infrastructure investment over the 2026 to 2030 period.
Now, where is China now? And where does that compare? Because 9,800 exaflops, maybe that, you know, just sounds like Chinese to all of you out there listening, but it is like the measure of computational power aggregated. And in July, that same ministry in China reported that China's intelligent computing capacity that's relevant to AI applications had reached an aggregate of 2,185 exaflops at the end of June. That's up 177% over the year before. So, it's growing quite rapidly. By contrast, America right now has 40,000 exaflops.
So, America has more than four times the amount of compute that China expects to have by the year 2030 and more than 20 times the amount of compute that China has today. And you know, how much is America going to have in 2030? Well, you know, there's various scenarios, but China is more than doubling every year. And so far, America has been doubling or close to doubling the amount of computing infrastructure for AI that's going in the ground thus far. And it's plausible. There are various scenarios put out by Epoch AI and others that foresee America having 250,000 exaflops by 2030 or 500,000 exaflops by 2030.
So, it's not like the export controls have accomplished nothing. There is this massive compute advantage that I don't think would be there in an alternate universe where these export controls had not happened. At the same time, there really is this gap between how much compute do you need to run all the experiments that give you the recipe for creating a frontier-level model versus what is the actual amount of compute you need to train such a model.
Once you have that recipe, and the combination of industrial espionage to understand what are all the tricks that are being used by the AI companies, what are the results of their experiments, along with distillation, which is to extract the results of that recipe in the form of useful training data — that means that what took the American companies gigawatts of compute and months of time might actually only take the Chinese companies, which have far less compute, hundreds of megawatts and months of time. Because like the actual training run, and this was disclosed by Dylan Patel of SemiAnalysis, but like the actual training run of training a Mythos might be like a 100-megawatt or 200-megawatt data center running continuously for a month. And there are Chinese AI labs that have that much compute.
What they don't have is the other gigawatts of compute required to run all the experiments that let them know that that training run is going to work, or the gigawatts that serve the hundreds of millions of customers on the inference side. So distillation is really critical in making up that gap, as is industrial espionage. So that to me explains why Xi Jinping still doesn't like these export controls, why he would still really want them gone, because of that 500,000 exaflop versus 9,800 exaflop that might be the gap by 2030. But it also explains why he's not desperate, why he's not coming groveling, begging for compute. He feels like he has a decently strong hand.
The second way you mentioned China improving its AI posture is by building up its own domestic chipmaking and equipment-making capacity. Compute self-sufficiency is something Beijing has promoted since at least Made in China 2025, which was released in 2015. And national champions like Huawei have been the standard bearers for advancing this kind of self-sufficiency push over the last decade or so. How is that effort coming along heading into these talks?
Well, I think it still remains a priority and something that they're working incredibly hard on. I think one thing that struck me as quite interesting is that we've been talking about these five-year plans that are coming out that are sector by sector. So, another five-year plan, the electronic information manufacturing five-year plan, which was issued jointly by MIIT, that ministry we were just talking about, as well as the NDRC, came out on September 15th, and it lists full-chain breakthrough in integrated circuits as its number one priority. So what they're saying is, of all the things that we have to do in the electronics industry, our top goal is full-chain, meaning full supply chain breakthrough in integrated circuits.
I mean, that's listed above building AI data centers. That's listed above leapfrogging the United States in AI leadership. I mean, what they really care about is are they self-sufficient in semiconductor technology? And it makes sense. I mean, if you're China, you're importing more by dollar value of semiconductors every year than you are oil. Even with all the export controls, that's still true. And that is a terrifying thing to potentially lose access to, and explains why they're so focused as something that they need to make progress on. And here you're seeing evidence that even though they're still behind, even though they're still not where they want to be, they are still making progress.
So, one piece of recent news on this topic that struck my eye was a September 4th piece by McKenzie Hawkins in Bloomberg where she reported that DeepSeek, the leading Chinese AI model developer on many metrics at least, is planning to install a new data center full of 160,000 Huawei chips and 1 gigawatt of energy capacity going to be built out in Inner Mongolia. Now, according to the report, these are going to be Ascend 950DT chips, which is the latest generation of Huawei's AI chips. It is generally assumed to be roughly comparable to an Nvidia Hopper generation of chips. So, for Nvidia, those were the chips that they announced in early 2023. Huawei announced the 950D in 2025 and appears to actually be making some of them in 2026.
When are they going to have 160,000 of them? Not at all clear. The article says that it's going to take more than a year to fulfill this order. That makes sense based on some of the other details it includes, such as Huawei is planning on producing 1.6 million Ascend dies across its various production network, which mostly includes SMIC's advanced node manufacturing capacity, but also perhaps some of the capacity that Huawei itself is bringing online internally. And 1.6 million Ascend dies is enough for about 600,000 910C chips, which uses two of these dies, and then you have to factor in yield issues. So not a ton of 950DT production expected to take place this year, which means they might not be fulfilling this order for a very long period of time.
In fact, just six months ago, it was a breakthrough when one company brought online a compute cluster that used 10,000 Huawei chips. And then in June, somebody was reporting the first ever model trained on 50,000, quote, 'domestic Chinese chips,' although we don't really know if those were even made in China or merely just designed in China. But all this is to say that even though China is way behind in the compute capacity story, they're still making progress. They're still better this year than they were the year before, and they're still better than many, many countries around the world.
I think the last point of comparison would just be SpaceX AI. When they built the first iteration of the Colossus supercluster in the second half of 2024, that was 100,000 Nvidia Hopper chips. So that means sometime in the 2027-2028 timeframe, DeepSeek might have a data center equivalent on domestic chips to what SpaceX had in 2024. Still a pretty significant gap, especially when you take into account that by that time companies like SpaceX AI are going to be deploying Vera Rubin chips or maybe even what's the next generation of Nvidia chips, which are just way better on a per-chip basis than what Huawei's coming out with.
And Greg, in the AI value chain, chips are just part of the picture. You wrote a Substack in August about how China is making a significant push to produce semiconductor manufacturing equipment domestically in order to reduce the reliance on companies that are more aligned with the US in both East Asia and Europe. How is China doing on that front?
Well, so that Substack piece was really about lithography specifically, which is arguably the most important piece of equipment in a chip manufacturing operation. Although, you know, which is more important, your brain or your heart? You need both of them to live. But Chinese suppliers have been pushing to domesticate semiconductor manufacturing equipment for a long time. They had explicit metric targets in Made in China 2025, that policy we talked about from 2015. But in general, there is a group of companies that are making progress on this, denoted as just advancing their own technological sophistication and capacity of production.
So in July when we were writing that Substack piece, Shanghai SiCarrier, which I'm sure I'm pronouncing that wrong, Electronic Technology Group reportedly began volume production of immersion DUV machines. And we're talking like single-digit numbers this year, double-digit numbers next year. In some sense, you could say that puts them two generations behind ASML. So ASML right now is making what's called high-NA EUV machines, like extreme ultraviolet lithography, and then since 2010 ASML has been making extreme ultraviolet lithography. The most advanced type of equipment before that was this immersion deep ultraviolet lithography technology, which was already mature in ASML in the early 2000s.
But Huawei, even though it is best known here in the United States as a telecom company, even though they're also pretty well known now for their AI chip part of the story, they are a big player in the semiconductor equipment space, both internally as a user of it and also because they have subsidiaries. So, there was a great piece in the Financial Times on September 7th by Ellie Alcott, and what they wrote is, quote, 'Huawei is seeking to close the gaps by backing local suppliers and helping push their equipment into commercial fabs where years of testing and refinement are needed before domestic machines can compete with established foreign equipment.' The other thing Huawei is doing is that they are backing different component suppliers.
They're backing domestic Chinese competitors to Zeiss like Fujian Gichi Photonics and Gigaphoton, competitors for light sources like Beijing Rofin Laser Optoelectronics. So they're really trying to, as that explicit government policy goal states, to build the entire value chain on an all-Chinese basis. Sort of top priority is de-Americanization, but right behind that is de-foreignization.
So, US export controls, at least according to an analyst cited in that piece, could accelerate Chinese progress. Quote, 'If the restrictions push local fabs to collaborate with local DUV suppliers, they could increase their technological capabilities much faster than expected.' End quote. That's from Lin Qingyuan, an analyst at Bernstein, which is a major equity analyst shop that does semiconductor equipment research among many other parts of the semiconductor value chain.
I would just say that the causality could also run the other way. If SiCarrier and others were selling directly to Huawei, if Zeiss was selling directly to Huawei, yeah, they would be de-Americanizing much faster because whatever the current pieces of the puzzle that they can't make indigenously, they can patch that near-term with foreign sources as part of the long-term plan to make everything domestic Chinese. So selling to these companies can actually accelerate their indigenization if you give them critical components that they currently can't make earlier than they would otherwise be able to. And in fact that's exactly what's going on right now.
If you look at the fabs inside, for example, YMTC, you know, they talk about what share of the equipment is Chinese, but that's at the OEM systems integrator level. If you break open those machines, they still have a lot of critical foreign content inside. And that appears to be the case with these immersion DUV machines that are going on in China as well. They're still critically reliant on foreign components. If we were able to effectively cut them off from that access to components, then they would not be in a position to start testing inside Chinese fabs to begin that years-long process of refinement that's required for commercial adoption.
So, I just don't inherently agree with the idea that export controls accelerate indigenization. It kind of depends on how you implement it, which is why I've been so frustrated with the half-hearted implementation, because I think in many ways that does accelerate domestic indigenization.
Okay. Okay, so coming back to the reasons why Beijing is feeling good heading into this summit. We first have their illegal and legal access of American compute, and then you also have on top of that their ongoing efforts to boost domestic chip capacity. Let's talk about the third one you laid out at the onset of this episode, which was Chinese open models. In recent months, a flurry of more advanced models have come out of China's labs that are getting closer to the frontier set by US models, and China has been looking to lean on their open AI models to push a global alternative to the US's closed model ecosystem. How has that strategy been playing out for them, Greg?
Well, I think it's still the case that the United States is unambiguously setting the pace. I thought Treasury Secretary Scott Bessent said something that was quite interesting the other week in which he said China is effectively copying off of America's homework, and you can't get a higher test score than the person you're cheating off of. You're sort of limited by how much knowledge they have as the maximum amount of knowledge that you can use to get a good test score. So, it's definitely the case that the best models right now like GPT-6, Astra, are meaningfully better than the best stuff that China has.
It's also the case, I think, that to the extent that there has been reporting that Chinese models are better than American models, that's using benchmarks that are often saturated, which is to say the nature of model performance is poorly captured by these benchmarks because of the way the frontier is shaping up. And Chinese model producers, they often even say in their own industry trade press when they're criticizing their competitors, 'Oh, those guys aren't actually as good as they claim to be.' You know, they're just trying to saturate the benchmark.
So, on the one hand, they're clearly good. They're clearly delivering economic value for customers who care about getting that kind of value. On the other hand, they're not pushing the frontier forward the way the American companies are, and they're maybe not even as good as they say they are when they're emphasizing all these various benchmarks. That said, they're good enough to drive adoption, and they're still unambiguously the leaders in the open-weight ecosystem.
And there is a universe out there that really cares how good the best open-weight models are. And we saw a really interesting case study in this on September 3rd when Saudi Arabia's sovereign AI company Humain announced a Chinese-built frontier model after it entered into a joint venture agreement with the United States in 2025 for advanced chips. And I think this is such a remarkable story because we all remember a year or two ago when it was a hugely contentious policy debate in the United States of like, should we be selling chips to the United Arab Emirates? Should we be selling chips to Saudi Arabia? And the debate really did hinge upon, well, are they going to choose Team America or are they going to choose Team China?
If selling them chips can lead them to durably attach themselves to Team America, then that makes selling the chips much more obviously straightforward as a good idea. But if we're going to sell them all these chips and greatly empower them, and then their loyalty to America is still highly uncertain and they're keeping a foot in both camps of America and China, then it's a much more debatable proposition as to whether or not selling them these chips is a good idea.
Let's just talk about who Humain is. We called them the sovereign AI company, but it's really important to note that they were launched in 2025 with the Public Investment Fund of Saudi Arabia's money, which is chaired by the crown prince of Saudi Arabia. Here's a quote from the press release from the Public Investment Fund explaining what Humain is. Quote, 'Humain was strategically launched in May 2025 under the Public Investment Fund to build the entire AI stack: data centers, cloud infrastructure, models, and applications, chaired by His Royal Highness Crown Prince Mohammed bin Salman. The company is poised to be a significant contributor to Saudi Arabia's Vision 2030.'
So the US, when they're dealing with Humain, it's not like it's some Wild West kind of company defying the wishes of the Saudi government. This company is for all intents and purposes an extension of the Saudi government. So the chips that they got were approved in November 2025, and the two countries signed a strategic artificial intelligence partnership which the Commerce Department heralded. That allowed them to buy tens of thousands of Nvidia Blackwell chips, the best chips that you can currently buy on the open market.
Now that Humain is actually at the stage where it's ready to start developing and releasing its own models, a very fascinating thing has taken place. Humain commissioned MiniMax, a Chinese company, to build an Arabic model based on its own Chinese open-weight model. So, Humain M3, announced at LEAP in Riyadh on September 3rd, is a 428-billion-parameter mixture-of-experts model built based on the MiniMax M3 lineage and further pre-trained on more than a trillion tokens of Arabic language content.
So, Humain asserts that it is not training a Chinese model but extending access. And according to the CEO of Humain in an interview with Semaphore, he's specifically talking about the agreement that the United States and Saudi Arabia reached when that chip deal was struck. And he's saying, quote, 'This is all acceptable under the framework with the United States because Humain is not furnishing a training cluster to Chinese frontier labs. We are only providing inferencing.' And then he goes on to say, 'This is not about importing a model and putting a Humain label on it. Humain defined the benchmarks, commissioned the model, drove the Arabic optimization, and has integrated it into Humain's broader intelligence and infrastructure stack.'
I don't think that's the strongest defense I've ever heard. Number one, commissioned the model. Who is doing the training? It is a Chinese company doing the training. They did provide this trillion tokens of Arabic language content, but MiniMax trained the first version of the model, and they're overseeing all the technical work to train this Arabic-derived version of the model.
Now, what we've learned based on these remarks is that the way the agreement was structured is that you're not allowed to do training of Chinese models, which is debatable based on even the facts we have right now whether or not that is in fact what they're doing. But they're claiming that, oh, it's only going to be on the inferencing side of the equation, and that makes it true. So if that's true, if the agreement didn't block inferencing models, you still have to ask yourself what are the geopolitical implications of this.
And here I'm thinking about the fact that the vast majority of Chinese models, when they are served around the world, are served by American cloud companies. And that's just because Americans have the vast majority of global cloud computing power. And that's true pretty much everywhere around the world that you might actually go get cloud computing availability. Now, if the United States wanted to stop allowing Chinese open-weight models available on its cloud, that right now would be devastating to the Chinese model developers in terms of global adoption.
But this Humain story illustrates how that might not always be true. If we were to hypothetically try and block Google, Microsoft, AWS, Oracle, and others from serving Chinese models on their cloud, well, we don't want to just be telling everybody, hey, if you want to do Chinese models, who you actually need to buy from is Humain. And so, this is the tension. The more we sell our chips abroad, the more we might be tying our hands or restricting our freedom of movement in other ways that we might try to counterbalance China later.
This is a really tough set of issues to wrestle with. If you're the United States, you face a really tough set of trade-offs. And I thought what Michael Kratsios, the director of the White House Office of Science and Technology Policy, said at the India AI Impact Summit back in February — he was really promoting AI sovereignty as a reason to partner with America. You know, buy American AI hardware, train your own sovereign AI models, own your own sovereign AI data centers, and you can do whatever you want with it. But what if what you want to do with it is to train models that are origin Chinese or provide Chinese models around the world?
This is an interconnected supply chain, and it's really, really tough to cut China out of one place if you're not willing to try and cut them out of every place. And the way we apparently struck the deal with Saudi Arabia did not envision keeping our options open this way going forward.
Thanks, Greg. I think that's a great example of the way that Chinese open models have been adopted globally. But it's also worth remembering that here in the United States, there's also examples of Chinese models being adopted by American companies. There have been numerous reports over the summer as these Chinese models have been coming out that they are cheaper and capable enough to be adopted by the likes of people like Airbnb, DoorDash, and countless startups. So this spread of Chinese open models is not just one that happens internationally, but it also happens right here in the United States.
And you got to imagine Xi Jinping loves every one of those headlines.
And real quick, Greg, just to wrap it up, the fourth reason you mentioned Beijing is feeling good heading into these talks was about distillation. Can you just give us a quick preview of that?
Yeah, we've already covered this multiple times on the podcast. The thing that I do want to harp on is this recent US intelligence community report on distillation, because there's a lot of meaty new examples that I think make it pretty obvious that Chinese companies are getting a non-trivial plus-up. I mean, the way I've described it in the past is, of course, China is good because they have a lot of smart people, but why are they noticeably better than American open-source companies like Meta, for example? And I think distillation goes a long way towards explaining that gap. It explains so much of how China has been able to stay so close to the frontier despite all the limitations that they're facing. We'll touch more on that IC report later in this podcast and just go through it more specifically. So folks should definitely stick around for that.
Great. That's a good overview of the way that China feels good walking into these talks. I can imagine though that it's not all good news. What is it about AI, the pace of AI or AI development, that's worrying Beijing heading into these talks?
Yeah, I mean, I think there's three basic things. The first is the US export controls still bite. China is making progress relative to where they are, but they're still a fraction of where they would be in a world where these export controls never had happened. Right now, China is aiming for, as we mentioned earlier, 9,800 exaflops by 2030. America is aiming at 250 to 500,000 exaflops by 2030. That is a massive and growing disparity. In a world where the export controls didn't exist, it's totally plausible that America and China would be neck and neck in compute capacity, and totally plausible that China would be ahead in terms of setting the pace of the frontier. So that's the first thing.
I think the second thing is these AI-enabled cyber attacks are starting to get scary, and they're starting to get scary to senior officials in China's security state. They're saying, 'Oh my gosh, you know, if this is what's happening when America is accidentally hacking itself, what might be happening intentionally?'
And then the third thing, and this is something where I think we now have enough data points to say that opinions in the Chinese regime are changing on this topic, which to be fair, it's also true that opinions have massively shifted in the United States leadership. And that is on the topic of humanity losing control of AI systems. So it's not a one-to-one how China sees this, how America sees this. But I do think both countries have seen their opinions meaningfully shift in the past six months, even really since Mythos and the Hugging Face incident.
Yeah, let's dig into this a little bit more. Both the United States and China are heading into these talks amidst a flurry of calls to slow down the pace of development for fear of AI risks. We covered that on the pod last week. But this time, Chinese models are drawing closer to American ones in terms of performance, and the US knows it probably can't fully pace the frontier without Chinese models agreeing to do the same. That begs the question: does China see AI competition this way? Do they view AI safety and AI risks the same way that the United States does?
