In the second half of the 20th century, the race to develop nuclear weapons occupied some of the greatest minds in the United States and the Soviet Union.
Now, the United States finds itself engaged in a different kind of race with a different adversary: China. The goal is to master technology, specifically artificial intelligence (AI).
It’s a battle being waged in research labs, university campuses and the offices of cutting-edge startups, under the watchful eye of the leaders of some of the world’s richest companies and the highest levels of government.
The cost runs into trillions of US dollars.
Each side has its strengths, something Cut Wright, a cognitive neuroscience researcher at the University of London (UCL), aptly summarizes as the battle between “brains” and “bodies.”
Traditionally, the United States has led the field of AI: the world of chatbots, microchips, and Large Language Models (LLMs).
China has been superior in developing AI “bodies” – robots (and, in particular, “humanoid” robots that look uncannily like people).
But now, with both sides eager to prevent their rival from dominating, those advantages might not last forever, and the race could transform even further in the coming years.
The battle for dominance of Large Language Models (LLM)
On November 30, 2022, Californian technology company OpenAI launched its new chatbot. In a six-sentence statement, the company announced that it had trained a new model “that interacts conversationally.”
It was called ChatGPT. Immediately, the tech world was dazzled.
“You could go on any social network and find a flood of posts from people talking about all the different ways they were using this new text box that had appeared on the Internet,” says Parmy Olson, columnist for Bloomberg and author of Supremacy: AI, ChatGPT, and the dart that can change the arena.
It was the birth of the first Large Language Model (LLM) in widespread use.
An LLM analyzes large amounts of text and data that already exist on the Internet and uses it to learn patterns in the way recommendations are expressed.
Now, experts agree that when it comes to so-called AI “brains,” the United States leads the way.
OpenAI claims that more than 900 million people use ChatGPT weekly, almost one in eight people on the planet.
Other American technology companies, such as Anthropic, Google and Perplexity, were quick to keep pace, investing billions of dollars in creating rival LLM systems.
These AI companies know that, if they do it right, LLM systems can begin to take on many of the roles currently performed by humans in white-collar professions, and that commercial success translates into big profits.
How did the Americans play their cards?
But another question is also being asked in Washington: how will all this affect the United States’ race with China for global supremacy?
According to a senior US official who spoke to the BBC, the key to America’s strategic advantage lies less in extraordinary algorithmic programming and more in the hardware that powers the immense processing power: in particular, microchips.
Simply put, most of the world’s high-end, high-powered computer chips—those used by Silicon Valley companies to power the creation of machine learning machines—are controlled by the United States.
In fact, most of them are designed by a California-based company: Nvidia.
In October, Nvidia became the first company in the world to reach a valuation of $5 trillion. It could be the most valuable company of all time, according to Stephen Witt, author of The Weighting Machine.
And Washington uses a strict network of export controls to prevent China from appropriating those powerful chips.
This policy dates back to the 1950s, when the United States blocked exports of advanced electronics to countries allied to the Soviet Union.
And it was notably reinforced in 2022, under the presidency of Joe Biden, as the race for artificial intelligence intensified.
The United States can exert its influence on export controls, even though most of those powerful chips are not even made in the United States.
In fact, many are manufactured in Taiwan (a US ally), by the Taiwan Semiconductor Manufacturing Corporation.
The United States is ensuring that very few of those high-end chips made in Taiwan end up in China.
It does this through its “foreign direct products rule,” which requires foreign companies to comply with U.S. standards if the products they export contain U.S. components or are derived from U.S. technology.
The Taiwanese microchip factory is almost visible from mainland China. It’s easy to understand why the island could be a tempting target for Beijing.
So why don’t Chinese factories start producing those powerful chips themselves? It’s not that simple.
To make high-end chips, a UV printing machine is needed. Only one company in the world makes such machines: ASML, based in a small town in the Netherlands.
The United States uses the same tactic (its “foreign direct goods rule”) to prevent that Dutch company from sending those valuable machines to China.
This protectionist policy seemed to have been quite successful in helping the United States maintain its advantage in the field of artificial intelligence.
But now, China counterattacked.
DeepSeek’s counterattack
In January 2025, the same week that Donald Trump became president for the second time, surrounded by billionaire tech tycoons, China launched its own AI chatbot: DeepSeek.
For the user, the experience is very similar to that of ChatGPT. You can answer questions, write code, and it’s free to use.
Interestingly, DeepSeek is estimated to have cost a fraction of what it took to create American AI systems like ChatGPT and Claude.
It caused quite a stir. On January 27, 2025, Nvidia suffered the largest single-day loss in market value in the history of the US stock market: around $600 billion.
“It was extremely disconcerting for Washington,” says Karen Hao, a journalist specializing in AI.
She believes that the US policy of export sustainability may have been counterproductive: Chinese developers had to do without powerful chips, forcing them to be creative.
“Ultimately, this accelerated China’s self-reliance,” he says.
The distinctive feature of DeepSeek is that, at the time, it had similar capabilities to American models such as OpenAI and Anthropic, but using a much smaller number of chips for its training.
In Beijing, meanwhile, there was palpable optimism, says Selina Xu, a researcher who works on AI policy in China in the office of former Google chief Eric Schmidt.
“Everyone was trying to figure out how DeepSeek did it. And that has really been a very positive catalyst for the Chinese AI ecosystem.”
It also revealed a marked difference in the way the two countries operate.
In the United States, AI companies jealously protect their intellectual property, while China has taken a more open approach.
To accelerate adoption and innovation, Chinese companies often publish their code online, allowing developers at other companies to consult it.
“This means that Chinese technology companies, when creating a new AI model, do not have to start from scratch,” explains Olson. “They can just take that model, develop it and improve it.”
As a result, the race for AI “brains” is no longer so clear.
The United States believed that LLMs were a powerful tool in its arsenal; Now, China can make them too.
“American closed-source models are probably better, but maybe not by much,” says Selina Xu. “The Chinese model is maybe only 90% as good, but it is 10% more expensive.”
China’s advantage in the robotics war
And when it comes to AI “bodies”—the world of drones and robotics—China has historically had the upper hand.
Since the 2010s, the Chinese government has dramatically increased its support for robotics development. It funded research and provided robot makers with billions of dollars in subsidies.
It is estimated that there are now around two million robots in operation in China, more than in the rest of the world combined.
Olson says much of this success is because China is a manufacturing economy. “They have all that experience in electronics manufacturing, they take advantage of it and incredible robotics startups emerge.”
International visitors to Shenzhen or Shanghai are often surprised by the deep integration of robots into everyday life, Xu says; for example, home food deliveries with drones.
China has especially excelled in so-called “humanoid” robots: machines designed to look and behave like people.
The Center for Strategic and International Studies (CIS), a bipartisan American research center, has reported on a “dark factory” for cars in Chongqing, in the south of the country.
The plant has 2,000 robots and autonomous vehicles that they claim can deliver a new car every minute. It is called a dark factory because it is fully automated and, in theory, can operate in the dark without human presence.
Beijing is aware of the country’s rapidly aging population, Xu says. The government believes humanoids can fill the gap left by the retirement of human workers, especially in the care sector.
“It is expected that by 2035, the number of people [en China] people over 60 years of age exceed the total population of the United States,” he adds.
Not only is China building robots for its huge population, it also currently accounts for 90% of all humanoid robot exports.
The ghost in the machine
But there is a drawback.
China leads the world in building robotic bodies. However, each of those bodies still needs a brain: an operating system, or tool, that tells the different metal parts what to do.
If the robot only has to do a repetitive task, like the one it might perform at the Chongqing car factory, it only needs a relatively simple robotic brain. China can make it itself.
But for a robot to perform many varied and complex tasks, it needs an intelligent brain powered by a different form of AI, called agentic AI.
This is a program and AI that behaves more like an independent agent, executing tasks that consist of multiple steps.
So when it comes to those high-performing brains, the United States still has the advantage.
“The United States… is definitely still in the lead when it comes to robotic brains,” says Wright, the UCL researcher.
“Those are the AI chips and tools that help the robot perform specific tasks. And what you have to keep in mind is that approximately 80% of the value of a robot stays in its brain.”
About robotic dogs and drones
Both the United States and China are competing to combine robots with agentic AI, and one American company has shown that it is no longer just Chinese companies that can create successful robots.
And who wins matters: It’s a technology that could be both exciting and terrifying.
Boston Dynamics, an American engineering company, already uses it. His dog-shaped robotic, Dwelling, has become an online icon among tech fans, with millions of views on YouTube.
The robotic dog has powerful “eyes” (a high-tech camera with thermal imaging) and “ears” (acoustic monitoring).
Dwelling can now perform inspections in company warehouses, detecting problems such as equipment overheating, leaks or gasoline spills, before sending that information to the industrial AI tool provider, IFS.
The AI analyzes the results and makes decisions—possibly without human intervention—to solve the problem.
On the other hand, Wright states that we can already see the combination of robotics and AI on the battlefield.
Last summer, Ukraine began deploying the Gogol-M, a mother drone capable of flying hundreds of kilometers into Russia before launching two smaller attack drones.
Without human sustain watch over, these drones used their AI to scan the terrain, determine targets, steer toward them, and detonate explosives.
Who will win?
It’s difficult to predict who will emerge victorious when we don’t know the final outcome, says Greg Slabaugh, professor of synthetic vision and AI at Queen Mary University of London.
“It is astonishing that ‘victory’ is a specific moment, like the moon landing,” he adds.
“Instead, what matters is sustained advantage: who leads in capability, who integrates AI most effectively into their economy, and who sets global standards.”
With technologies like electricity and computing, Professor Slabaugh explains that it mattered less who built the systems first and more who deployed them most effectively across the economy: “The same thing could happen with AI.”
We don’t know where AI will take us. The large American technology companies want to launch themselves into that uncertain future without restrictions; The Chinese Communist Party, on the other hand, wants the state to supervise the investigation.
One vision promises a hyperactive version of consumer capitalism; the other, a world where the State determines what can or cannot be done with this technology.
“Each side has a better chance of winning on its own turf,” says Mari Sako of Oxford University’s Said Business School.
“When two players compete with different rules, I suspect that the one seeking to attract a broader audience—users, adopters, etc.—is more likely to prevail.”
And there is a lot at stake. It is still unclear whether the United States or China will emerge stronger from the 21st century. The race for AI could well be decisive.
Additional information: Ben Carter
click hereto find more stories from BBC News World.
You can also follow us on YouTube, instagram, TikTok, x, Facebookand in our new whatsapp channelwhere you will find breaking news and our best content.
And remember that you can receive notifications in our app. Download the latest version and activate them.
- 6 expert tips on how to talk to AI to get better answers
- I hacked ChatGPT and Google AI in 20 minutes (and what I managed to prove with that)
