Monday, November 16, 2026
Newsletter About
Opinion

In the Race for AI, China Is Falling Behind

The Nvidia scandal promises to compound the country’s disadvantages.

This article was originally published by The Atlantic and is republished here under license.

Nathan Chen, a teenage tech prodigy, was a leading innovator at Moonshot AI, one of several top Chinese labs that have sparked fears in Silicon Valley that China is gaining ground in the race for AI supremacy. But last month, Chen posted on X that he had left the Beijing-based company. He is now at Discovery Loop, a new Palo Alto–based AI start-up with a growing stable of high-profile talent.

Perhaps Chen sensed that the most exciting advances in AI were not happening in China—despite the fact that the technology’s rollout in the United States has lately looked more troubled than ever. Newly empowered Democrats, ushered to midterm victories thanks in part to public opposition to data centers, are calling to rein in construction plans. A sex scandal, possibly also a spy scandal, at Nvidia—a U.S. chipmaker that has almost single-handedly enabled the AI boom—has cost Jensen Huang, its former chief executive, his job; fomented turmoil in U.S. financial markets; and plunged the entire AI economy into crisis.

But China still appears to be falling behind. Chinese AI firms are short of chips they need—a problem that Huang’s downfall might well make worse—and they can’t raise enough capital to compete with their wealthy American rivals. Public resistance to the potentially job-killing technology seems to be growing, and the security-obsessed Communist Party, which has been bullish on the potential for AI to fix China’s mounting economic challenges, has lately appeared wary of the threats it might pose to the party’s grip on power. This attitude shift could further hamper AI’s progress in the country.

Proponents of the emerging technology argue that AI will bring societal benefits—boosting economic productivity, fast-tracking scientific advances, and strengthening military capabilities. China’s leaders have touted the technology as crucial to the country’s continued rise, and tech executives and national-security experts in the United States have framed the pursuit of AI as an existential race for technological dominance, a winner-takes-all competition between the two countries that will determine the global balance of power. Some analysts question whether the geopolitics of AI are so neatly zero-sum, but if advances do end up conferring a decisive strategic advantage to the country that best harnesses these tools, then China’s leaders should be worried.

Anthropic’s initial public offering, scheduled for this week, demonstrates the kind of access to capital that American firms enjoy and their Chinese competitors can only dream of. Anthropic is expected to raise more than $90 billion from its Nasdaq debut and direct sale of stock to investors, including Google and Microsoft. By comparison, a Chinese start-up known as Z.ai raised $558 million as the country’s first generative-AI firm on the Hong Kong Stock Exchange, in January, and an additional $9 billion in two subsequent share offerings. The story is the same when it comes to paying customers. Anthropic posted revenues of more than $16 billion in the first half of 2026, compared with Z.ai’s $142 million. The business model preferred by Chinese firms will make it hard to close this gap. Most offer their standard models open-source, which means the tech is free to use and update, and sell access to their most advanced models. But in China’s crowded AI sector, fierce competition for customers is driving down fees. Earlier this month, the Chinese tech giant Alibaba tapped Li Jiaqi, a livestreamer and beauty influencer known as the “Lipstick King,” to pitch a temporary promotion in monthly membership fees for the company’s AI productivity platform, from 28 yuan down to .01 yuan—or .0015 cents.

[Matteo Wong: A $5.7 trillion face-plant]

Chinese firms are also at a serious disadvantage in computing power. Washington has imposed export restrictions on Nvidia’s most advanced chips to China, and China’s own chipmakers have yet to make a product that matches their power. Chinese companies have tried to evade these controls by tapping cloud services and data centers outside the country and smuggling in banned Nvidia chips. But Huang’s extramarital affair with an alleged Chinese spy will likely lead to the closure of at least some of these loopholes. Alice Huang, the woman at the center of the Nvidia scandal (no relation), was a top executive at Megaspeed, a Singapore-based cloud-computing firm that has been under investigation by the U.S. federal government for illegally funneling Nvidia chips to China. Lawmakers from both parties have announced plans in recent days to add measures tightening restrictions on chip exports to the defense package currently stalled in Congress.

Chinese labs have demonstrated that they are able to build competitive AI models with less money and computing power than their U.S. rivals. But the advantages that America’s AI sector has in funds, infrastructure, and chips promise to compound over time. Chinese companies appear to have responded to this resource gap by devoting their energies to building lower-cost products for a wider user base, rather than trying to make the most intelligent cutting-edge models. “I don’t actually think at any point over the past two years that Chinese companies have been trying to beat the U.S. on intelligence,” Kendra Schaefer, a partner at the research firm Trivium China who specializes in Chinese AI, told me. “It has not been part of the strategy at all, because they know they can’t win.”

Chinese AI engineers have instead set their sights on where they can aggressively compete. For example, China’s massive manufacturing base would seem to grant an advantage in developing embodied AI for operating robots and other hardware. But here, too, technology from America is leading the way. Last month, Wang Xingxing, the CEO of Unitree, a top Chinese developer of humanoid robots, gushed about the capabilities of OpenAI’s newly released ChatGPT-6 Astra model for advancing the movements and actions of intelligent machines.

China might also seem to have an edge over the U.S. when it comes to the speed with which individuals and companies are adopting the technology. The Chinese government has an ambitious plan to spread AI through every aspect of its economy and society by 2035, and surveys find that Chinese people are far more optimistic than Americans about AI. This has raised concerns in the U.S. that China is doing a better job of realizing gains from the new technology. But Martin Chorzempa, a senior fellow at the Peterson Institute for International Economics, has found that in practice, American companies are integrating AI into their operations more extensively than Chinese companies, which means they “are able to get a lot more out of it,” he told me. He added that there is “zero” evidence that AI is spreading faster and deeper in China.

China’s leaders seem to have also grown warier of AI and the risks it could pose to the country’s political and economic stability. Beijing’s spy chief, Chen Yixin, recently called for greater regulation of AI to prevent China’s adversaries from using the technology to spread misinformation or to launch cyberattacks on critical Chinese infrastructure. But Chinese authorities appear just as nervous about more prosaic uses of the technology.

A recent editorial in the People’s Daily, the Communist Party’s mouthpiece, criticized some popular AI-generated videos that imagined a romance between two beloved characters from classic Chinese literature as “malicious alterations that ridicule heroes.” Chinese social-media companies duly scrubbed the short films from their platforms—in keeping with a wider government ban on AI-generated content that, in the eyes of the state, distorts Chinese history and culture and threatens to manipulate public opinion. In a country where the government seeks to control the national story, even the most innocuous AI creations can seem ominously subversive and worth suppressing.

Public anxiety about AI’s effect on jobs is also spreading, which may pose some obstacles on Chinese companies that hope to use the technology to increase productivity and cut costs. One of the country’s most popular chili-sauce brands, Lao Gan Ma, came under fire last month after the company’s chief quality officer credited AI with reducing product-defect rates and increasing revenues. These comments quickly circulated online, along with reports that the technology had allowed Lao Gan Ma to lay off hundreds of workers—a claim that the company has denied. Some consumers have responded by calling for a boycott of the sauce. As one commenter wrote: “Their layoff is our future.”

[Liza Lin: China has no use for AI doomerism]

Despite efforts to encourage companies to adopt the technology, China’s leaders are also clearly worried about related job losses. The government released new guidelines last month for beefing up job training and support for workers displaced by AI. These new provisions also allow regulators to weigh the potential job losses from AI-related ventures that require state approval, such as an expansion of robotaxis, suggesting that the government may intervene to prevent large-scale layoffs. AI is meant to help companies become more productive by allowing them to cut staff, yet China appears willing to sacrifice some of these benefits if it means preserving the social order. In the U.S., the fragmented regulatory landscape for AI gives companies, developers, and employers a great deal of latitude for exploiting the technology’s benefits.  

Perhaps the very idea that China and the United States are engaged in a high-stakes race for AI dominance misunderstands the situation. China might simply be claiming a role in AI that the country has taken in so many other industries: as a low-cost producer, offering sufficiently capable services at cheap prices that complement rather than rival the superior products sold for higher fees in the U.S. As in other industries, the demand for AI products at both the high and low end should be strong.

But if China’s government continues to block American AI models from operating in the country, these seemingly symbiotic market niches won’t end up serving the Chinese people, who will be largely doomed to the exclusive use of their country’s inferior products. If the Communist Party seeks to control what Chinese AI companies get to use and how they get to use it, then the United States will enjoy a monopoly not only on superior AI but also on the freedom to make the most of it.

More in Opinion

View All →

Leave a Reply