Elon Musk, Nvidia's Jensen Huang, Sam Altman, Satya Nadella and Google co-founder Sergey Brin sat at the same table last week inside the White House — an odd sight, given how often they compete for the same investors and the same chips. The occasion was rare too: Xi Jinping's first state visit to the United States in more than a decade.
Artificial intelligence dominated the dinner conversation. Xi told Trump that Beijing and Washington now have more room for cooperation than competition, and proposed working together to keep the technology from being misused. AI, he said, should stay under human control and serve progress rather than become another flashpoint for mutual suspicion. Trump responded that both sides should keep talking and deepen cooperation specifically on AI.
The most concrete idea came from US Treasury Secretary Scott Bessent, who discussed setting up an AI incident alert system with Chinese Vice Premier He Lifeng — something like the hotlines nuclear powers use to defuse false alarms. If a model in one country goes haywire, triggers a major outage or turns up in a cyberattack, the other side would hear about it directly instead of finding out from the news.
Alongside the AI talk, Washington and Beijing extended their trade truce by another two months, into early January. It's the second high-level meeting of 2026: Trump flew to Beijing in the spring, and now Xi has returned the visit. The friendly tone doesn't erase the fact that export limits on Nvidia's latest chips remain in place, and both countries keep racing to build their own training capacity.
It's still unclear what the alert-system idea will actually become — there's no treaty or even a formal memorandum yet. But the fact that AI risk made it onto a state-visit agenda, rather than staying confined to technical working groups, suggests both governments now treat it as a real concern, even as neither shows any sign of easing off the race for AI supremacy.
