China is winning one AI race, the US another – but either might pull ahead

China is winning one AI race, the US another – but either might pull ahead

In the latter half of the 20th century, the global race for nuclear supremacy captivated the strategic minds of the United States and the Soviet Union. Today, the United States faces a new competitor in the AI arena: China. The stakes have shifted from weapons of mass destruction to technological dominance, particularly in artificial intelligence. This battle is waged across research facilities, academic institutions, and innovative startups. It is overseen by leaders of major corporations and top-tier government officials, with budgets stretching into the trillions. Each side possesses distinct strengths, as noted by Nick Wright, a cognitive neuroscience researcher at University College London, who frames the contest as a clash between “brains” and “bodies.”

The US has historically held an edge in AI brains—chatbots, microchips, and large language models (LLMs)—while China has excelled in AI bodies, notably humanoid robots that closely resemble humans. However, as both nations strive to outpace each other, these advantages may no longer be decisive. The AI race is evolving rapidly, with potential shifts in the near future. In late 2022, a pivotal moment occurred when OpenAI, a California-based firm, unveiled ChatGPT. With a six-sentence announcement, the company revealed a model capable of conversational interaction, sparking widespread excitement in the tech community.

“You could go on any sort of social network and there was just this flood of posts from people talking about all the different ways that they were using this new little text box that had appeared on the internet,” says Parmy Olson, Bloomberg columnist and author of *Supremacy: AI, ChatGPT, and the race that will change the world*.

ChatGPT marked the emergence of the first widely adopted large language model. These systems analyze vast amounts of existing text and data to recognize patterns in how ideas are communicated. Experts now largely agree that the US leads in this domain, with ChatGPT boasting over 900 million weekly users—nearly one in eight of the world’s population. Competitors like Anthropic, Google, and Perplexity swiftly followed, investing billions to develop their own LLMs. Their goal is clear: AI systems could soon perform tasks previously reserved for human professionals, and success in this field promises significant financial rewards.

Yet, Washington’s focus extends beyond software to hardware. A senior US official tells the BBC that the country’s strategic edge hinges less on algorithmic innovation and more on microchip technology. High-end chips, essential for powering LLMs, are predominantly controlled by American firms. Nvidia, a California-based company, dominates this space, becoming the first to surpass a $5tn valuation in October. According to Stephen Witt, author of *The Thinking Machine*, Nvidia may well be the most valuable company in history. To secure this advantage, the US enforces strict export controls, a policy rooted in the 1950s but intensified in 2022 under President Joe Biden.

Despite the fact that most advanced microchips aren’t manufactured in the US, American influence persists. Many are produced in Taiwan, a US ally, by the Taiwan Semiconductor Manufacturing Corporation. Washington ensures these chips don’t easily reach China through its “foreign direct product rule,” which compels foreign firms using US technology to comply with US regulations. The proximity of Taiwan’s semiconductor plants to mainland China underscores the island’s strategic importance. Could Beijing one day claim this critical resource? For now, Chinese manufacturers struggle to produce such high-end chips, making the US’s control of the supply chain a vital asset in the global AI rivalry.