Is China Winning the A.I. Race?

I spent 20 years reporting from more than a dozen countries, and it occurred to me one day, what kind of newsletter would I like to read? I don't live in the US. I want something especially for a global audience. The world is just that. Each weekday morning, we bring you the biggest stories, dispatches from my colleagues on the ground, and a few surprises with video too.

The World's newsletter, read the latest and sign up at nytimes.com/theworld. From The New York Times, I'm Natalie Ketroff. This is The Daily.

President Trump is preparing to make a crucial trip to China this week to meet with

Chinese President Xi Jinping. A key question hanging over that meeting is artificial intelligence, and whether the global race in AI may be spinning out of control. Today, my colleague Vivian Wong explains how China's approaching artificial intelligence differently from the United States, and what that means about who's winning the global competition for AI dominance. It's Monday, May 11th.

“Vivian, we've come to you because you've been reporting on China for six years, is that right?”

Yes. Okay, and your beat has essentially been AI of late, and I think there is one main question

that everyone is asking when it comes to the global AI race. And that is is China beating us?

I think that's a really hard question to answer because China and the United States are actually running very different races with different goals and different metrics of success. And I think China is very confident that, along its chosen path, it is doing very well. Okay, so what is that chosen path? Talk to me about that. From the beginning, China's strategy when it comes to AI has been putting it in people's hands,

putting it in factories, putting it everywhere throughout the economies. This focus on real-world

“applications. And that's really different from the way that I think Silicon Valley and a lot of”

American policy makers talk about AI, which is generally revolving around AGI, right? This idea that AI is building towards this super powerful, superhuman intelligence, whereas China and Chinese

policy makers have always kind of talked about AI as this way towards making the economy run better.

So, if you just walk around on the street in China, you are going to encounter different manifestations of AI everywhere. There are driverless cars in a ton of Chinese cities. There are robots everywhere from restaurants to just patrolling the streets to definitely, of course, in factories. And people have found ways to use it in places that we might never have imagined. I recently talked to a few different parents who have thought of really creative ways to use AI

in their children's educations. I like myself. One mom had bought this AI contraption that was basically a translation mask that she would wear to help her kids practice speaking English. A translation mask. Yeah, it really looks like the mask that they in wears in the Batman movies. Mommy, who do you like? And the way it works is that she would speak into it in Chinese, and then what would come out was a version of her voice but speaking English.

I like you. Mommy, add out with you. Wow. And so she would just wear it around the house to talk to her kids. What are you eating? I'm making a pumpkin right now. Is it tasty? Yes, it's a delicious.

“You have to really love AI to put on a bean mask to talk to your kid. I mean, I just can't”

totally imagine it, but that's commitment. Yeah, I think that what's been really striking to me about observing just the way that people in China are talking about and thinking about AI is people are really excited about it. You know, in America, when people talk about AI, there's this whole doomer conversation. A lot of people are really nervous about it, you know, either taking their job or taking over the entire world. Right. And that conversation is just a

lot quieter in China. Okay. So when you say that China feels like it's doing pretty well, given the path that it's taken on AI, what you mean by that is making people kind of fall in love a little bit with AI. It sounds like

to them now. And that comes in part because China has a lot of really deep structural issues that

without AI, I think they really don't have a way to solve. So for example, the population and the workforce are aging really quickly. And how do you solve that if young people don't want to have children? Well, maybe using AI to make your factories more efficient. Another problem is huge inequities between access to healthcare in rural and urban areas. If you don't have enough doctors, how do you solve that? Maybe AI doctors are the answer. So it sounds like in a sense, there may be quite a bit

writing on AI from the Chinese government's perspective. If it's being seen as a potential

“solution to all these really important problems. I think so. Yeah, the Chinese government historically”

has always been quite optimistic about technology. I mean, if you look at the past few decades,

technology has been a big part of what has powered China's amazing economic rise. And so I think they see AI is basically the latest iteration of some kind of technology that is going to help them on this global extent, and is going to solve these problems that they have. Got it. So when does that relationship between the Chinese government and AI actually begin? What's that story? So a key moment is in 2014. Well, what could you find out in the direction? It was early in Xi Jinping's tenure as

leader of China, and he was giving a speech to a gathering of some of China's top engineers and scientists. And he had gotten a bunch of reports from the scientists ahead of time, kind of giving him suggestions

“on what, you know, they thought China's next direction in science should be. And she gets up there,”

and he gives a speech, and he says, "Oh, thank you so much for all of your suggestions. I read them all." And I was really interested in this one report that said, "intelligent robots are going to be a really big deal going forward." And he basically says, "I want China to be on top of that. I want China to be the number one maker of intelligent robots." And I also want to sediment all sorts of other technologies that are going to be the future. And this turns into a plan in 2017 that is called

China's New Generation AI Development Plan, which is their first national level roadmap of what they are going to do with AI. And it sets out a goal that says they want China to be a leading AI power by 2030. And it also says that the way it's going to get there is through real-world applications. And given how things typically go in China from understanding, I'm assuming that what happens next

is what always happens when they have a plan, which is that they pour boatloads of money into it

into supercharging it. Am I right? Yeah, exactly. It kind of sets the years of government going at all levels. So after this national plan comes out, then you have the provincial and local governments sort of falling over themselves to show how they are complying. Yong-jiao is attempting to recreate a Silicon Valley style work environment. Embagings Yong-juan soon district. So they start building industrial parks to attract top companies. They start giving tax breaks and sort of free or cheap

“rent to these companies. It was a good decision to come back to China. I think it's easier to”

get funding here than Silicon Valley. And they start giving them money to attract the top talent. It's really this all in national effort to deliver. And what's the effect of that? So for a few years, China really feels like it has the lead in AI. Police in Beijing have been wearing these glasses that can recognize faces linked to the government's national data base. It can tell whether it is an adult, a child, a male, or female. How advanced is this technology?

This is a cutting edge. Especially at the time, they were really focused on developing technologies like facial recognition and voice recognition. And Chinese companies really were at the forefront of this. They were some of the best in the world. And this was, of course, also really useful for China's surveillance apparatus. We're trying to transform our economic model from Made in China to create it in China. But then in 2022, Chad G.P.T. comes on the scene.

Chad G.P.T. Chad G.P.T. is technology accessible and free to anyone on the web that impersonates what it's read on the internet. This promises to be the viral sensation that could have completely reset how we do things. And that really caught Chinese policymakers and Chinese tech companies off guard. It caught everybody off guard. Yeah, exactly. And for China, I think it was a moment that made them sit back and think,

completely different area, generative AI that they had not been thinking about as much. And it

“kind of scares the Chinese government because generative AI, you know, it's the thing that is looking”

at all of the information on the internet. It's coming up with answers that are sort of unpredictable. And the Chinese government becomes really nervous that this is going to be a threat to their information controls. Their censorship regime. You know, they don't want Chinese people plugging questions about politically sensitive events into a chatbot and getting answers that they don't want the people to be getting. Hmm. So they're looking at this as a potential threat to their

political model. Exactly. I think that, you know, they had been thinking about AI as this really

powerful tool for the economy and in some ways to solidify their grip on political control.

But here is proof that it also could be a problem. And so you see them react. ChatGPT is banned in China. The Chinese government starts putting out new regulations also on Chinese AI companies that might be working on generative AI. For example, they quickly roll out a rule that says, "If any Chinese AI company is building a model that can, quote unquote, mobilize society, it has to first pass a series of checks and clearances with the Chinese government before it

can be released to the public." Hmm. What are those checks and clearances? So basically,

“you have to do safety testing. But I think when we think about safety testing in the West, right?”

It's making sure that the chatbot is not feeding its users answers that might push them to self-harm or, you know, feeding hate speech. In China, there is that element. But there is also the element of making sure that it doesn't provide answers about politically sensitive questions like the June 4th massacre in Tiananmen Square or about China's top leaders that the government might not want Chinese people asking. It sounds like there's this tension for the Chinese

government between wanting to dominate in AI, which now includes this world of generative AI. But also from an authoritarian perspective, wanting to maintain as much control as they can over information, which is so key to powering these models. I mean, part of the reason for the success of these chatbots is that they pull from the entire universe of information. Yeah. And this is a tension that exists in China's entire economic development,

where the government knows that to allow its economy to flourish, it needs openness, it needs interaction with the rest of the world. But of course, it also wants that interaction to happen on its own terms and within its own control. So that tension is just even more pronounced when it comes to AI. And so they are trying to maintain that balance, trying to figure out how that works. And then in early 2025, along comes Deepseek. Right. Let's talk about Deepseek,

because it is mind-blowing and it is shaking this entire industry to its core, the emergency. So Deepseek was a chatbot created by a Chinese startup that really no one had heard of before last year. And it performed basically just as well as leading models in the United States.

Deepseek says that their AI model only cost 5.6 million dollars. No, we don't.

But according to the company's founders, it had been trained for so much cheaper. It was open source. And to have the Chinese actually open sourcing it, meaning that that is going to be what goes around world, huge, huge step-up and change, and puts us a little bit on our back foot in trying to figure. And it was this wake-up call in this sign to everybody around the world that Chinese AI could compete

“with Silicon Valley. I think Mark Andreason put it really smartly when he said that this”

feels like the sputnik moment for the AI race. And for all of China and the Chinese government in particular, this is also a moment of huge national pride. Soon after Deepseek's release, Xi Jinping actually called a meeting of a bunch of business leaders, including the founder of Deepseek,

who's just a guy in his 30s, again, who most people have never heard of before this point.

And Xi tells the entrepreneurs there, go out and show your talents and serve the country. And this is seen as a really important signal that the government recognized that it needed to give companies a little bit more space if it really wanted them to be on the cutting edge. And so I think what it's essentially trying to do here is have its cake and eat it too. I talked to an analyst who I thought put it really smartly. He said the Chinese government is trying

to control what AI says, but unleash what it does. And so basically, if you are making a chatbot,

people, you are going to be under a greater degree of restrictions. But if you are doing robotics, if you are doing something that doesn't really deal with information that could directly influence how Chinese people are thinking, there are a lot fewer restrictions there. And did that meeting, did the new signaling, new messaging around this, actually lead to concrete changes

“that you see in China, like do you see it manifest? I think so. I mean, it's hard to say if it came”

directly from that meeting or just the excitement around deep-seeking general. But 2025 and this year have really just been huge moments of development in China's AI scene. You've seen a ton of new models getting registered. You've seen a bunch of startups doing really well. And that's also where you've seen kind of AI get diffused so broadly into the economy in all the ways that we talked about or everyone's using it all the time. Right. And we see how much they're doubling down on

AI just this past March when the Chinese government released its five-year plan, which is this grand economic plan for how China's economy is going to run for the next five years. And this plan

mentions AI more than 50 times. Wow. So it's really clear that they are staking basically their

entire economic future on dominating AI. And for Washington, that puts China on a direct collision course with US tech. We'll be right back. I'm Jackson Jones. I'm a reporter and meteorologist at the New York Times. For about two decades, I've been covering extreme weather, which is getting worse because of climate change. And it's becoming more important to get timely and accurate weather

“information. That's why we send these customized newsletters letting you know up to three days in”

advance about extreme weather that could impact you or a place you care about. At the times, you can be confident that everything we publish is based off the most accurate, scientific and vetant information available to us because we want you to be able to make real-time decisions about how to go about your life. This is the kind of work that makes subscribing to the New York Times so valuable and it's how you can support fact-based, independent journalism. So if you'd like

to subscribe, go to NYTimes.com/subscrib. So in the collision course that the USC's China running headlong into, what are China's disadvantages? I wondered, for example, if some of the restrictions that the government still has on the development of AI, the control that it exercises over it might hold their industry back in

“some meaningful ways. Sure, and this is really the million dollar question, right? And I think the answer”

is both yes and no. So when you look at just the models themselves, the deep seek for example, experts I've spoken to actually say that they don't think the restrictions are really making deep

seek meaningfully less powerful. In the sense that sure there are some topics that deep seek cannot

answer, but actually chat about some other countries also have red lines or sensitive areas that they can't talk about. They just might be different. And so when it comes to actually just the core performance of the model, if you're not asking about these sort of few politically sensitive questions, experts seem to think that actually China is not meaningfully held back. Now on the other hand, I think there is a potential issue when you look at the broader political environment,

and the reality that the government has shown that it's willing to crack down at any point if it feels threatened or a little bit not in control of the entire situation. And I think that creates a really uncertain business environment that could make it difficult for Chinese companies to thrive. Just talk about what the effect of that could be. It sounds like it could have a chilling effect, but what is the impact of knowing that the government could clamp down at any moment?

I think it just creates a sort of weight in the back of the entire Chinese private sectors

mind at all times. They always have to be thinking about where are the red lines that we have to

be careful not to cross. And part of the issue is that that's not always clear. Sure, there are certain really obvious ones like the question of Taiwanese independence, but it's really hard to know what might trigger a new red line or new dissatisfaction by the Chinese government. And also when it comes to interacting with the rest of the world, Chinese companies and founders have to be really careful to not to maybe get investment from the wrong people or the wrong countries.

We actually saw a really clear example of this just recently.

after it claimed to be the world's first general AI agent. The agent is able to execute complex tasks like market research, coding, and data analysis. It eventually moved to Singapore, but it's founders were Chinese, it's parent companies, Chinese, and last year Meta closing out 2025 with an 11th hour acquisition. The company announcing it is buying Manus.

Meta announced that it was going to acquire Manus for $2 billion, which was a really big deal,

but pretty much immediately after this was announced, the Chinese government said that it was investigating this acquisition. And just last month, China has blocked Meta's $2 billion takeover of AI startup Manus. They actually announced that they were not going to let the deal go forward. This comes as Beijing continues to take steps to restrict US investment in developing Chinese technologies. And on top of that, during all of this, during this investigation, my colleagues

reported that Manus's founders, after they went back to China for a visit, were actually blocked from leaving China. What you're talking about with this case is really the Chinese government actively curtailing the ideal environment that you would have for innovation. And it's interesting because China is famous for doing all of this aggressive industrial policy making to fuel its industries at home. But it sounds like AI may just be a different beast.

Yeah, AI is just such a new field and also such a diverse field, right? There are so many different types of AI. And so the government can't just say, okay, this is the particular

“direction that you need to go. We're going to throw money at this particular problem and we will”

suddenly be the world's AI leaders. You really need to give scientific minds and researchers kind of the space and freedom to experiment and to kind of follow where they're scientific intuitions or their research is leading them. There's a long-standing stereotype that might be a bit of an over-generalization, but a lot of people tell me that they think it still holds true, which is that China is really good at the one-to-ten of scientific research, which is taking

some kind of scientific discovery and finding the ways to just apply it everywhere and really make it

useful. But it's not as good at the zero-to-one, which is just getting to that breakthrough

discovery in the first place. And I think that's where the political environment does play a little bit of a role. Right, you can't force innovation from a top-down strategy. So is that the mean disadvantage that China sees itself having vis-a-vis the US or their others? There are two other really major areas. So one is chips and another one is talent. So on chips, which are the powerful

“hardware that's necessary for training AI models, the fact is that Chinese chips are just not as powerful”

as American ones. So Chinese AI companies are still really reliant on chips from Nvidia to make the best AI that they can. Nvidia, an American company. Yeah. And China has spent hundreds of billions

of dollars on trying to develop domestic powerful chips, but they just haven't gotten there yet.

And then the other problem is talent. So China produces many, many, many extremely talented AI researchers every year. But without the sort of open innovative environment that we were just talking about, many of them may choose and do choose to go to the United States. And so Chinese companies and Chinese policymakers have talked very openly about how a shortage of talent is another problem that they have to face. These are pretty significant handicaps that China has. Just to

summarize what you've said, it's an imperfect environment for innovation, less powerful chips, and this talent issue. But at the same time, we've also spent the better part of this conversation talking about all the ways in which China is ahead in making AI practically useful in people's lives and making sure it is everywhere. So just to return to that question that we wrestled with

“in the beginning of this, is China coming out on top in the global AI race? I think if the end”

destination of all of this AI is indeed AGI, this superhuman intelligence that is just going around the world, then I do think American companies are closer to that. I do think Silicon Valley is closer to the cutting edge of technology. But it is important to point out that China is not that far behind. It's a few months behind by most experts estimates. But if AGI is not the final destination, I mean, if it's not even possible, and the real way to go is to just use AI to improve our daily

lives now, then I think China has the edge because it already has deployed AI so much more than the

cars, it's also generating a bunch of new data all the time that is making these Chinese models

even better every single day, which helps solve one of the bigger problems in AI these days, which is kind of just running out of training data. China is just constantly generating new data

“for its own models. So at the end of the day, I think that the problem is AGI is still so full of”

so many unknowns. We don't know what the future of this technology is. We don't know what exactly it means to dominate AI or what it would take to dominate AI. And so it's really impossible to say who is pursuing the right path. Vivian, do you think there's a lesson for US companies in China's approach to AI? Because it does seem as though the ambition of reaching toward the superhuman

intelligence is so quintessentially American. It's the biggest most innovative version of this tech

possible, but it's also potentially backfiring in a sense because it seems to be turning a lot of people against this technology very forcefully, whereas focusing on more mundane, maybe less ambitious goals may be a more popular route. Yeah, I think it is a really interesting contrast with the United States where the public hesitation or even pushback to AI is creating in many ways political problems for these AI companies. But in China, because people are feeling that this AI is actually

tangibly improving their lives, it's making their commute smoother, it's making their food get delivered to them faster, it's making products maybe cheaper because factories are more efficient,

“people really feel that this technology could be something that is good for them. And I think”

it's worth noting that we have started to see the conversation in China shift a little bit, especially in recent months. We have seen people start to talk a little bit more about what happens to my job if AI becomes everywhere. But I think that sentiment is still not as maybe prominent as it is in the United States. And actually that is maybe an interesting result of the government restrictions that we talked about. So yes, on the one hand, they might make it more difficult for

Chinese companies to innovate. But on the other hand, I think there's a general sense among many Chinese people that the Chinese government will keep this whole sector under control. It will make sure that AI is just not getting out of hand and leaving society totally unstable. And so that also creates a certain degree of trust in the technology. Right. And I think ironically in the US, the fact that we have this more free-wheeling environment when it comes to AI, there's something

about that that makes people more afraid of it. And that backlash from the grassroots right now, that couldn't be a problem down the line for AI development in the US. That could be a headwind. Yeah, I think that could prove to be a real impediment to AI development in America, that is probably just not going to play out in the same way in China. Well, Vivian, thank you so much. Thank you, Natalie.

We'll be right back. I'm Kevin Rus. I'm Casey Newton, and we're the host of Hard Fork, a show from the New York Times about technology and the future. About the future that's already here, Kevin, every week on the show, we bring your news from the front lines of tech interviews with key newsmakers, wacky experiments that we get up to. And we just generally have a lot of fun. Yes, so whether you're curious about

developments in AI or just what's happening on TikTok, we are here for you. So that's hard for you can find it wherever you get your podcasts.

“Here's what else you need to know today. So what the Virginia Supreme Court did was not only wrong,”

it was disgraceful. They basically... Democrats are scrambling to respond to a Virginia Supreme

Court decision on Friday, which struck down a new congressional map that would have given the party an advantage in the midterms. You know, the voters will have to last word on this. The fallout of the decision is huge. Even though Democrats are still expected to win the house, they now have to flip more seats in hostile territory. In its fourth or three decision, the courts said that Democrats had violated the state's constitution by drawing a new map meant to

give the party 10 of the state's 11 house seats. In a private call on Saturday, Democratic lawmakers discussed trying to restore the map with a move that would replace the entire state Supreme Court

drastic solution. And, passengers and crew from the cruise ship that was hit by a deadly

“hand-to-virus outbreak were evacuated on Sunday. The ship anchored off the Canary Islands in Spain”

were health officials checked the passengers' temperatures and symptoms. The Spanish Health Minister

said everyone on board was asymptomatic. Today's episode was produced by Shannon Lin and Rikina

“Vetsky. It was edited by Lisa Chow, and contains music by Marion Lazzano, Pat McCusker,”

and Dan Powell. Our theme music is by Wonderley. This episode was engineered by Alyssa Moxley.

That's it for The Daily. I'm Natalie Kittrella. See you tomorrow.