Artificial Intelligence Has Become a Battlefield Between the US and China, and Nobody Knows Who Will Win
Artificial intelligence has become the new battlefield between the two greatest powers on the planet, and this showdown is far from having a definitive winner.
If the 20th century had the world holding its breath over the arms race between the US and the Soviet Union, the scenario today is different, but the tension is just the same. This time, the American adversary is China, and the weapons have changed: algorithms, microchips, language models, and increasingly sophisticated robots. The difference is that instead of missiles and nuclear warheads, what is being measured now is the ability to process data, train massive models, and put intelligent machines to work at an industrial scale.
This race is playing out in research labs, university campuses, and the offices of cutting-edge startups, all under the watchful eyes of leaders from some of the wealthiest companies on Earth and the highest ranks of both governments. The cost? Trillions of dollars. And each side has its advantages, at least for now.
Nick Wright, a cognitive neuroscience researcher at University College London, sums up what is at stake with a spot-on analogy: it is a battle between brains and bodies. On one side, the US dominated the world of LLMs for years, the large language models behind tools like ChatGPT, along with the chatbots and microchips that power all of it. On the other, China quietly built one of the largest fleets of robots on the planet, with a particular emphasis on so-called humanoid robots, machines designed to look and act like people. But that clean divide over who leads where is starting to crack. With the emergence of DeepSeek in early 2025, China showed it can build AI brains too. And while Americans are still trying to figure out how that happened, the race keeps accelerating, with each side trying to cross into the other’s territory before it is too late. 🤖
The Brains of the Battle: LLMs and the American Dominance That Started to Shake
On November 30, 2022, OpenAI, a company based in California, launched a new chatbot with a press release of just six sentences, announcing it had trained a model capable of conversational interaction. The name was ChatGPT, and the tech world was immediately awestruck.
As Parmy Olson, Bloomberg columnist and author of the book Supremacy: AI, ChatGPT, and the race that will change the world, described it, all you had to do was open any social media platform to see a flood of posts from people talking about every possible way to use that new little text box that had appeared on the internet.
That was the birth of the first major large language model aimed at a mainstream audience. An LLM analyzes massive amounts of text and data already on the internet and uses them to learn patterns in how ideas are expressed. OpenAI says more than 900 million people use ChatGPT every week, nearly one in every eight people on the planet. Other American companies like Anthropic, Google, and Perplexity scrambled to keep up, spending billions of dollars building rival systems.
For many years, when it came to generative artificial intelligence and large language models, the game seemed pretty much settled. American companies built the most powerful LLMs in the world, fueled by billions of dollars in investment, unrestricted access to top-tier hardware, and a robust academic ecosystem. The narrative was comfortable for the American side: whoever has the best chips and the biggest investment wins. These companies knew that if they got it right, LLMs could take over many of the functions in white-collar jobs that humans perform today, and that commercial victory would mean a whole lot of money.
The Strategic Role of Microchips in the American Advantage
But the minds in Washington were focused on a different question: how would all of this affect the US race with China for global primacy? According to an American official who spoke with the BBC, the key to the strategic advantage of the United States lies less in extraordinary algorithmic coding and more in the hardware that drives the immense computing power, particularly in microchips.
Most of the high-performance chips used by Silicon Valley companies to power the creation of LLMs are controlled by the Americans. The majority of them are designed by a single California company: Nvidia. In October 2024, Nvidia became the first company in the world to be valued at 5 trillion dollars, possibly the most valuable company of all time, according to Stephen Witt, author of The Thinking Machine.
Washington uses a strict network of export controls to prevent China from accessing these powerful chips. This policy dates back to the 1950s, when the US blocked the export of advanced electronics to countries allied with the Soviets. But it was significantly tightened in 2022 by President Joe Biden as the AI race heated up.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The Americans can exercise this power even though most of the most advanced chips are not manufactured on American soil. In reality, a huge portion of them is produced in Taiwan, a US ally, by the Taiwan Semiconductor Manufacturing Corporation (TSMC). The United States ensures that very few of these cutting-edge Taiwanese chips reach China, using the so-called foreign direct product rule, which requires foreign companies to follow American rules when the exported goods contain US components or technology.
And chips are not the only bottleneck. To manufacture cutting-edge semiconductors, you need an extreme ultraviolet lithography machine. Only one company in the world produces these machines: ASML, based in a small city in the Netherlands. The US uses the same tactic to prevent ASML from shipping these machines to China. This protectionist policy seemed to be working well to help the Americans maintain their lead in AI brains. Until China hit back. 😬
The DeepSeek Counterpunch and the New Reality
In January 2025, the same week Donald Trump was inaugurated for the second time, surrounded by tech billionaires, China launched its own AI-powered chatbot: DeepSeek.
For the user, the experience is quite similar to ChatGPT. It answers questions, writes code, and is free. But what truly sent shockwaves through the market was the cost: it is estimated that DeepSeek was built for a fraction of the price required to create American LLMs like ChatGPT and Claude.
The impact was so significant that on January 27, 2025, Nvidia stock suffered the largest single-day loss of market value in the history of the American stock exchange: roughly 600 billion dollars vanished. The Chinese model showed it is possible to get around, at least partially, the dependency on the most advanced microchips by using more efficient optimization techniques.
Karen Hao, a journalist who covers AI, believes the American export control policy may have backfired. Chinese developers were forced to get by without the most powerful chips, which pushed them to get creative. The result, she says, was an acceleration of China’s self-sufficiency in the field of artificial intelligence.
The defining characteristic of DeepSeek is that it delivered capabilities similar, at that time, to those of American models from OpenAI and Anthropic, but using far fewer chips during training. In Beijing, the mood was one of palpable optimism, according to Selina Xu, a researcher who works on Chinese AI policy in the office of former Google CEO Eric Schmidt. Everyone wanted to understand how DeepSeek had pulled it off, and the model became a very positive catalyst for the Chinese AI ecosystem.
Open Source Versus Intellectual Property: Two Models on a Collision Course
The episode also highlighted a striking difference in how the two countries operate. In the US, AI companies fiercely protect their intellectual property. In China, there is a more open approach, with a focus on open source. To accelerate adoption and innovation, Chinese companies frequently publish their code online, allowing developers at other companies to study, adapt, and improve it.
As Parmy Olson explains, this means that tech companies in China, when building a new AI model, do not have to start from scratch. They can take an existing model and build on top of it. The result is that the race for AI brains is no longer so clear-cut. America thought LLMs were a powerful tool in its arsenal, but now China can produce them too.
Selina Xu puts it in perspective: the American closed-source models are probably still better, but perhaps not by as wide a margin. The Chinese model might be only 90 percent as good, but it costs just 10 percent of the price. And in many practical scenarios, that is more than enough.
The Bodies of the Battle: The Robot Fleet China Built in Silence
While the debate about LLMs dominated the tech world headlines, China was quietly building a robot infrastructure that is now the largest on the planet. Starting in the 2010s, the Chinese government dramatically ramped up support for robotic development, funding research and providing billions of dollars in subsidies to manufacturers. The current estimate is that there are roughly 2 million robots in operation in China, more than in the rest of the world combined.
Olson attributes much of that success to the fact that China is a manufacturing economy. All that expertise in building electronics was leveraged to create incredibly competitive robotics startups. International visitors who travel to Shenzhen or Shanghai are often surprised by how deeply robots are integrated into everyday life, from food deliveries by drones to autonomous robots that carry groceries right to your front door.
China has been especially prominent in so-called humanoid robots: machines designed to look and act like human beings. The Center for Strategic and International Studies, a bipartisan American think tank, documented the existence of a so-called dark factory in Chongqing, in the south of the country. The plant has 2,000 robots and autonomous vehicles that together can supposedly deliver a brand-new car every minute. It earned the name dark factory because it is fully automated and can, in theory, operate in the dark, without any human presence.
According to Selina Xu, Beijing is well aware of the rapid aging of the Chinese population and is betting that humanoid robots can fill the gap left by the retirement of human workers, especially in areas like elderly care. By around 2035, the number of people aged 60 or older in China is expected to surpass the entire population of the United States. Besides building robots to serve its own massive population, China today accounts for 90 percent of all global exports of humanoid robots. 🦾
The Ghost in the Machine: When Bodies Need Brains
China leads the world in building robotic bodies. But each of those bodies still needs a brain, an operating system or software that tells all those metal parts what to do.
If the robot only needs to perform a single repetitive task, like the ones in the car factory in Chongqing, a relatively simple brain is enough, and China can develop that in-house without any trouble. But for a robot to carry out varied and complex tasks, it needs a smart brain powered by a different form of AI called agentic AI. This is an artificial intelligence program that behaves like an independent actor, working through tasks that involve multiple steps.
When it comes to these high-performance brains, the Americans still have the edge. Wright, the UCL researcher, is emphatic: the United States is definitely still in the lead on robot brains. That includes both the chips and the AI software that helps the robot carry out real-world tasks. And here is an important detail: roughly 80 percent of a robot’s value is in its brain.
Robot Dogs, Drones, and the Frighteningly Near Future
Both the US and China are now racing to combine robots with agentic AI, and an American company has already shown that it is not only Chinese firms that can deliver successful robots. And it matters a lot who wins this stage: this is a technology that can be both thrilling and terrifying at the same time.
Boston Dynamics, an American engineering company, is already using this combination. Its dog-like robot, Spot, has become something of an internet icon among tech enthusiasts, racking up millions of views on YouTube. Spot has powerful eyes, a high-tech camera with thermal imaging, and ears, an acoustic monitoring system. It already performs inspections in the company’s warehouses, detecting issues like equipment overheating, gas leaks, or spills, and feeds that information into industrial AI systems that analyze the findings and make decisions, potentially without any human intervention.
On the more concerning side of the equation, Wright points out that you can already see the combination of robotics and agentic AI in another context: battlefield drones. In the summer of 2024, Ukraine began deploying the Gogol-M, a mothership-type aerial drone capable of flying hundreds of miles into Russian territory before releasing two smaller attack drones. Without any human control, these drones used their AI brains to scan the terrain, identify targets, and fly toward them to detonate explosives. It is the kind of advancement that makes anyone stop and think about the real implications of this technology.
Microchips: The Strategic Resource That Determines Who Moves Faster
At the center of this entire showdown are microchips. Without advanced semiconductors, there is no way to train large-scale LLMs, no way to equip robots with real-time processing capability, and no way to keep up with the pace of innovation this field demands.
The US understood this early and used its influence to restrict China’s access to the most advanced chips, especially after the export control measures of 2022. Nvidia, AMD, and TSMC were prohibited from fulfilling certain orders from China, a move widely interpreted as an attempt to create an insurmountable bottleneck in Chinese technological progress.
The problem is that the strategy worked partially but not completely. China responded on multiple fronts: it accelerated investments in SMIC, its main domestic semiconductor manufacturer, ramped up chip imports through alternative channels before restrictions were expanded, and, as the DeepSeek case demonstrated, began developing algorithms that squeeze more performance out of the microchips available. It is a parallel race where, at the same time the Americans try to turn off the faucets, the Chinese look for ways to need less water to make the same recipe work.
TSMC, based in Taiwan, is perhaps the most central player in this entire equation, even though it is often overlooked in more surface-level analysis. It manufactures the most advanced chips for Nvidia, Apple, AMD, and dozens of other companies. The fact that it is located in Taiwan, a territory over which China claims sovereignty, turns semiconductor geopolitics into something far more complex than a simple trade dispute. The Taiwanese fab itself is almost visible from the Chinese mainland, and it is easy to understand why the island could be a tempting prize for Beijing. That is precisely why the US has accelerated investments to encourage domestic semiconductor manufacturing on American soil, reducing dependence on a single island in the Pacific. 🖥️
Who Will Win This Race?
It is hard to predict who will win a race when nobody knows where the finish line is, as Greg Slabaugh, a professor of computer vision and AI at Queen Mary University of London, observes. According to him, the victory probably will not be a singular moment, like landing on the Moon. What matters is sustained advantage: who leads in capability, who incorporates AI most effectively across their entire economy, and who sets the global standards.
With technologies like electricity and computing, Slabaugh notes that it mattered less who built the systems first and more who deployed them most efficiently across the entire economy. The same may prove true for artificial intelligence.
There is also a fundamental philosophical difference between the two sides. The big American tech companies want to push forward into the unknown future of AI with as few restrictions as possible. The Chinese Communist Party wants the state to oversee that research. One version promises a hyper-edition of consumer capitalism. The other, a world in which the state determines what can and cannot be done with this technology.
Mari Sako, from the Said Business School at the University of Oxford, puts it well: each side is better positioned to win at its own game. And when two players compete under different rules, she suspects the player who wins the broader audience, the users and adopters, tends to prevail.
What Is at Stake Beyond Technology
It is tempting to see this dispute as nothing more than a competition between tech companies or a fight over market share, but what is truly at stake goes well beyond that. Leadership in artificial intelligence, LLMs, robots, and microchips will determine who has more productive capacity, who can automate critical processes, who develops more effective defense systems, and ultimately, who sets the standards and the rules that will govern these technologies in the future.
Historically, whoever sets the technological standard shapes much of what happens economically and politically in the decades that follow. The internet was American, the smartphone was American, and that had immense consequences for the global balance of power.
China clearly does not want to repeat the role of passive consumer of technology developed somewhere else. The Made in China 2025 plan, launched nearly a decade ago, already made clear the intention to dominate strategic sectors like robotics, semiconductors, and artificial intelligence. What has changed in recent years is the speed of execution and the quality of the results. DeepSeek is not an isolated fluke; it is a symptom of an ecosystem that has matured and now produces original innovation, not just adaptations of what was developed in the West.
On the American side, the response has been a combination of trade restrictions, massive public investments, and an effort to maintain alliances with partners like Japan, South Korea, the Netherlands, and the United Kingdom to create a kind of united front around semiconductors and AI. But alliances have costs, economic interests sometimes clash with strategic objectives, and China is too large a market for any tech company to simply ignore.
The stakes are high. It is still unclear whether the US or China will emerge more powerful over the course of the 21st century. The artificial intelligence race may very well be the deciding factor. And until that outcome becomes clear, every advance by one side quickly forces a response from the other, keeping the entire world watching every new move in this showdown that promises to shape the future of practically everything. 🌏
