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

During the latter half of the 20th century, the competition to create nuclear weapons became a focal point for the US and Soviet Union’s brightest minds. Today, the spotlight has shifted to a different contest: the race to shape the future of technology, particularly Artificial Intelligence (AI), where China and the US are vying for dominance. This battle is waged in laboratories, academic institutions, and the spaces of ambitious startups, with corporate titans and national leaders closely watching its progression. The financial stakes are measured in trillions of US dollars, and while both nations hold distinct advantages, these may not hold indefinitely.

Historically, the US has excelled in the realm of AI “brains” — the domain of chatbots, large language models (LLMs), and microchip innovation. China, on the other hand, has demonstrated greater prowess in the “bodies” of AI, such as robotic systems, especially humanoid robots that mimic human appearance with uncanny precision. However, as the competition intensifies, both sides are scrambling to prevent the other from gaining the upper hand, raising questions about whether these lead advantages will persist.

The Rise of Large Language Models

On 30 November 2022, OpenAI, based in California, unveiled its latest creation: a chatbot capable of engaging in conversational interactions. Named ChatGPT, the tool sparked immediate global attention. “You could go on any social network and there was just this flood of posts from people discussing how they were using this new text box,” notes Parmy Olson, Bloomberg columnist and author of *Supremacy: AI, ChatGPT, and the race that will change the world*. This marked the emergence of the first widely adopted large language model, which processes vast amounts of internet text and data to replicate the patterns of human communication.

Experts now concur that the US maintains a clear edge in AI “brains.” OpenAI reports that over 900 million users access ChatGPT weekly — nearly one in eight of the world’s population. American firms like Anthropic, Google, and Perplexity are investing heavily to match this progress, recognizing that successful AI models could soon automate tasks currently handled by white-collar workers. The commercial success of such systems is seen as a pathway to significant financial gains.

Strategic Control of AI Hardware

Yet, beyond software, a critical factor in the AI arms race is the hardware that powers it. A senior US official tells the BBC that America’s strategic edge lies not in the algorithms themselves, but in the microchips driving the computing infrastructure. These high-performance chips, essential for training LLMs, are predominantly controlled by the US, with Nvidia standing out as a key player. In October 2022, the company surpassed a $5tn valuation, potentially cementing its status as the most valuable entity in history, according to Stephen Witt, author of *The Thinking Machine*.

To sustain this advantage, the US enforces strict export controls, ensuring that China cannot easily acquire these advanced microchips. The policy, rooted in Cold War-era restrictions, was recently intensified by President Joe Biden as the AI rivalry escalated. Even though many of these chips are manufactured in Taiwan — a US ally — the nation’s influence extends through its “foreign direct product rule.” This regulation compels foreign companies to adhere to US standards if their products incorporate American technology or components.

Taiwan’s semiconductor factories are geographically close to mainland China, making them a strategic target. This proximity underscores the importance of securing access to cutting-edge hardware. Meanwhile, China’s challenge in replicating this technology independently remains significant. The production of high-end chips requires complex processes and specialized materials, which the nation is still working to master. As the race continues, the outcome could reshape the global balance of AI power in the years ahead.