As President Donald Trump meets with Chinese leader Xi Jinping this week in Beijing, the United States and China are locked in an intense competition to become the dominant player in the future of AI innovation.
Analysts suggest that artificial intelligence is the most transformative general technology since electricity brought light and power to average American households during the 1920s. And while today’s AI rivalry between the world’s two most powerful economies is already well underway, it could be nearing a critical juncture.
The AI race is accelerating. With both sides so focused on winning, they might be underestimating the risks the technology poses. And their competition may not be a true race since both contestants aim to cross different finish lines.
Why We Wrote This
In Beijing, President Donald Trump and President Xi Jinping are expected to discuss the future of AI, as both the U.S. and China compete to dominate this new technology. The outcome of the meeting – and the contest – could have a profound impact on the world’s two biggest economies.
“Both the U.S. and China are striving for global AI supremacy,” says Aynne Kokas, a professor at the University of Virginia who researches Sino-U.S. technology relations. “That being said, they’re approaching it radically differently.”
And while the current competition has echoes of the Cold War technology rivalries to build more and better nuclear bombs and to put a man on the moon, this competition holds the potential to drive far more consequential change – positive or negative – for humanity.
Each side, so far, is playing to its strengths.
Entrepreneurial America wants to maintain its qualitative advantage in AI and become the first to achieve artificial general intelligence (AGI), defined as machines or software that replicate human intelligence. Using advanced AI models, known as large language models (LLMs), the U.S. tech sector is striving to innovate and sell cutting-edge, world-beating products and services, from computerized office assistants to smart weapons.
Factory China, by contrast, is more concerned with integrating AI across every sector of its economy and society – from education to healthcare to government services and the military. Beijing also wants to bolster its global supply chains with AI and smart robots, so that it can remain the world’s most important exporter.
Julia Demaree Nikhinson/AP
“The great fear in Washington is that China gets to AGI first,” says Kyle Chan, an expert on China’s technology development at the Brookings Institution. “In China, the priority really is much more on these nuts-and-bolts applications.”
AI models and chips
What makes the competition so intense is that American and Chinese companies are struggling for control over the same AI tools, especially advanced AI models and the superfast chips to run them.
Washington has worked to keep those AI tools – especially the most advanced chips – out of Chinese hands. The U.S. does not want to give China’s already powerful manufacturing base any greater advantage than it already has for boosting exports to the U.S. and other countries. In its eyes, Beijing’s heavily subsidized, government-directed economy gives China an unfair advantage over Western companies.
John Locher/AP
For Beijing, the AI competition is “existential,” says Bill Drexel, a senior fellow at the Hudson Institute who researches AI competition with China. China’s leaders still chafe at what they see as a history of unequal trade and treaties imposed by the West, he says, “and they see technology as kind of the key ingredient that allowed that to happen.”
That is why China is prioritizing the development of homegrown AI infrastructure, free of dependence on the West’s software models or advanced chips. China’s technology sector has made important strides towards this goal in recent years.
A “Sputnik moment”
In January 2025, DeepSeek, a new and relatively unknown Chinese lab, released an LLM called R1, reportedly trained with less powerful chips and at a fraction of the cost of Western models. Nevertheless, R1 equaled, or nearly equaled, the performance of leading American LLMs in various tests. Some analysts referred to this as America’s AI “Sputnik moment.” The term refers to 1957, when the Soviet Union shocked the U.S. by launching its Sputnik 1 satellite, shaking public confidence in America’s technological superiority.
DeepSeek’s release sent U.S. tech stocks plunging. Shares of chipmakers Nvidia and Broadcom lost 17% of their value in a single day. Chinese AI companies have continued to release new, advanced LLMs. Last month, Stanford University’s highly regarded AI Index Report concluded that, with China’s DeepSeek-R1 rivaling U.S. top-tier models like Anthropic, the “U.S.-China AI model performance gap has effectively closed.”
But it’s important not to overdramatize China’s accomplishments, warns Jeffrey Ding, a political science professor at George Washington University and author of ChinAI, a weekly newsletter. China’s adoption of AI technology has been “shallow, narrow, and slow,” he writes in a recent edition. “These all-in-one machines have not diffused past early adopters and attracted few repeat customers,” Dr. Ding writes.
The U.S. retains some major advantages in AI development. “The U.S. still produces more top-tier AI models and higher-impact patents, while China leads in publication volume, citations, patent output, and industrial robot installations,” according to the Stanford AI Index report. On top of that, the U.S. also has more than 10 times as many data centers (where AI processing takes place) as any other country.
Still in early stages
Moreover, the AI race is still in its early stages. AI development isn’t leveling off, the Stanford report finds. “It is accelerating.”
At present, the U.S. is way ahead of China when it comes to capital expenditure by large tech companies, points out Dr. Ding. In 2020, U.S. companies were outspending their Chinese counterparts 6 to 1; by 2024, that advantage had grown to 10 to 1.
But Chinese state spending is helping narrow that gap. Still, U.S. government spending on AI rivals any federal investment in research and development since the height of the Space Race, according to some analysts. The combination of public money and burgeoning market investment in the U.S. may prove difficult for China to match.
Instead, China is investing heavily in two other pathways to advanced AI: neuroscience research and “embodied AI,” which integrates artificial intelligence into robots, drones, and self-driving vehicles.
This broader approach might prove successful, says William Hannas, lead analyst at the Center for Security and Emerging Technology at Georgetown University. “China is devoting substantial resources to the intersection of AI and neuroscience, and the embodied approach to AI is huge in China,” says Mr. Hannas, formerly the senior expert for China open-source analysis at the Central Intelligence Agency.
China’s goal is for AI embodied in robots and other machines to learn directly from environmental feedback – leading to improvements in performance and in the AI itself. “It’s more realistic,” Mr. Hannas says. “It’s the up-and-coming thing.”
Big productivity, or “race to the bottom’’?
The nation that most successfully incorporates AI at an industrial scale is likely to see the big productivity leaps that analysts expect down the road, but which have not yet materialized.
No one knows who will be the first to achieve AGI, a theoretical form of AI that can learn, reason, and apply knowledge in multidisciplinary settings.
One possibility is that the winner of this technology race will rule the world forever, Mr. Hannas says. Another is a rough parity, in which each country develops a unique form of AGI tailored to its own purposes. A third possibility is that advanced AI will pick its own winners or, possibly, leave no winners at all because one or both of the rivals neglected to put in safeguards to ensure the technology would not run amok.
U.S. officials reportedly have said that Mr. Trump and Mr. Xi will discuss AI security during their meeting in Beijing.
“Unfortunately, I think there is a bit of a race to the bottom in terms of safety, in terms of trying to address AI risks, because of the lack of trust between the two countries and the feeling on both sides that they need to be running faster,” says Mr. Chan of the Brookings Institution.