AI rivalry to dominate Trump-Xi summit as US and China spar over chips and models

The contest for artificial intelligence supremacy is expected to overshadow next week's meeting between U.S. President Donald Trump and Chinese President Xi Jinping in Washington, with the two powers still divided on access to advanced chips, allegations of technology theft and how the AI industry should be regulated.

Both governments treat progress in AI as central to their economic and military competitiveness. U.S. Treasury Secretary Scott Bessent is due to meet Chinese Vice Premier He Lifeng this weekend to discuss AI-related issues ahead of the leaders' summit.

Military risks

Security experts in both countries are pressing their governments to prepare for scenarios in which an AI system interferes with a nuclear command network or carries out a military cyber operation, leaving leaders with only minutes to judge whether an attack is under way. Proposed safeguards include red lines around nuclear systems, a requirement for human control over consequential cyberattacks, and a hotline for incidents involving autonomous AI. Among the unresolved questions are how to prevent an accident from being mistaken for an attack, which military decisions machines should never take, and how nuclear-armed rivals should communicate when technology behaves unexpectedly.

Technology copying

A technique known as distillation has become a flashpoint in the battle over AI model development. It involves training smaller AI models on the output of larger, more expensive ones to cut the cost of building a new tool. In September, the U.S. government accused six Chinese companies of maliciously copying American AI products this way, saying the firms acted "likely with Chinese government awareness." A similar accusation was made in April, before Trump's visit to Beijing. The Chinese embassy in Washington has called the allegations baseless.

Chip exports

Chinese access to the U.S. technologies underpinning the AI boom remains a source of friction. Since 2022, Washington has imposed sweeping restrictions on shipments of advanced chips and chipmaking equipment to China. There is evidence that advanced AI chips are being smuggled into the country, though the scale is unclear. The U.S. has also been weighing limits on the ability of Chinese firms to remotely access data centres in other countries, which would prevent them from benefiting from Nvidia's advanced chips. Some exceptions have been made: late last year, Trump decided to allow sales of Nvidia's H200 AI chips. The H200 is not the most sophisticated chip available, but the decision drew concern from China hawks in Washington who argued it could strengthen Beijing's AI capabilities.

Divergent approaches to regulation

The two leaders hold fundamentally different views on how far governments should involve themselves in the AI sector. Trump has dismissed calls for new federal AI rules, saying this week that the U.S. only needs a strong president, which he said the country has "in spades," and describing appeals from AI executives for regulation as a "hoax." China, by contrast, has rules governing AI algorithms and the data used to train them, including a requirement that AI-generated content be clearly labelled.

How far behind is China?

Estimates typically given by U.S. officials place Chinese AI models just a few months behind their American counterparts, though industry leaders increasingly describe the two as neck and neck. In 2025, then-White House AI adviser David Sacks said China was likely three to six months behind the United States. Earlier this year, the U.S. Commerce Department put DeepSeek V4 Pro about eight months behind frontier models. Many in the industry say the gap is closing quickly. Last month, Hugging Face Chief Executive Clement Delangue said China was winning the race in open-weight models and was probably not far off at the frontier either. "I wouldn't be surprised if they start dominating at the frontier either by the end of this year or next year at the rate of progress," he told CNBC on August 3. Frontier models are the most advanced AI tools, while open-weight models have publicly accessible core elements that can be downloaded, inspected and modified.