AI Supremacy Tops the Agenda as Trump and Xi Sit Down

Anthropic's top AI model leads the best Chinese system by just 2.7% as of March 2026, according to Stanford University researchers — a margin so thin it frames the stakes for Thursday's summit between US President Donald Trump and Chinese President Xi Jinping.

Few topics on the agenda will matter more to the global economy than the two countries' competing ambitions in artificial intelligence. Trump has been blunt about his position: "Whoever wins AI, WINS!" He has dismissed AI risk to humans as "a hoax" and argued the only guardrails the technology needs are a "high IQ" president.

Xi has not matched that rhetoric, but analysts see the same conviction in Beijing. "China very much sees it as a race, they very much think China should be the leading force in AI — they see AI as a power maximiser," says Rebecca Arcesati of the Mercitor Institute for China Studies. AI has anchored China's technological strategy for more than a decade.

China's foreign ministry pushed back on the framing, warning that "narratives of threat" and confrontation hinder global AI governance. It called for cooperation to ensure the technology benefits all, saying AI development was "fuelled" by the country's "vision of extensive consultation and joint contribution for shared benefit."

The scoreboard depends on the metric

American firms — Anthropic, Google and OpenAI — hold the lead in frontier models, the highest-performing systems on most benchmarks. In February 2025, a model from China's DeepSeek "briefly matched the top US model" before Anthropic pulled ahead again, Stanford researchers found.

The US advantage rests on two pillars. First, capital: frontier labs are burning through enormous sums to build training infrastructure, and a large share of US economic expansion now relies on the growth and projected profitability of AI companies. Second, export controls: sweeping US government restrictions bar American companies from selling advanced microchips to Chinese firms, chips that are crucial for cutting-edge development.

China's counter-strategy is scale, not frontier. Chinese developers release open weight models — downloadable and adaptable, unlike the proprietary systems US firms sell through subscriptions — at a rapid pace, with benchmarks showing the country's top models trailing US rivals by only a few months. Moonshot AI's Kimi K3 has gained traction in China and overseas by performing close to frontier models while remaining free.

"AI isn't the moon landing," said Lizzi Lee, a China analyst at the Asia Society Policy Institute. "The US may lead at the frontier, while China could still gain enormous economic and geopolitical leverage by making capable AI cheap, open and ubiquitous."

China also holds structural advantages. Its AI companies access electricity at much lower cost, Arcesati told Tech Life on the BBC World Service. In robotics, China is the world's biggest manufacturer, with more than two million robots working in its factories, according to the International Federation of Robotics. It accounts for more than nine-tenths of global humanoid robot shipments in the first half of the year, according to Chatham House.

Analyst Lian Jye Su of tech research firm Omdia said Chinese developers have shown "remarkable resilience and innovation" despite the chip shortage. US officials and AI firms counter that Chinese developers closed the gap through distillation — a technique that lets models learn from more advanced platforms.

Divergent rulebooks

Washington has prioritized innovation over regulation. "We're leading now over China by a lot, and everyone else, and we're going to keep it that way," Trump said this week, adding that the US would not "stifle the growth of something that will be bigger than the industrial revolution."

Beijing has moved earlier with centrally coordinated rules. Xi has urged leaders to keep AI's risks under watch and ensure the technology remains "under human control." China introduced guides on data usage and algorithms before many other countries, helping address underlying concerns, said Jayant Dave, information security officer at Check Point Software Technologies. Those guidelines are not legally binding but set expectations while leaving developers room to innovate, said Jonathan Sim, an AI ethics lecturer at the National University of Singapore.

Cooperation looks limited. The two countries cooperated on nuclear weapons because they shared an understanding of the threat — absent with AI. "Washington worries about capability escaping control. Beijing worries about capability threatening domestic stability," Dave said. Trump has rejected industry-wide slowdown proposals, arguing they would let Chinese developers catch up.

One concrete channel exists: ahead of the summit, US Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng discussed a notification mechanism for serious AI incidents. Policy analyst Leia Wang of the Carnegie Endowment for International Peace called AI's presence on the agenda a "good start" despite Trump's recently "more antagonistic" messaging — while warning against allowing an "unhealthy version of the race narrative to perpetuate."

For markets and AI firms on both sides, the summit's real test is whether a 2.7% margin becomes a firewall or a flashpoint.