In a hearing room on Capitol Hill last Tuesday, Nvidia's chief executive Jensen Huang sat down with Democratic and Republican leaders to make an unusual argument: that American companies should be allowed to keep using artificial intelligence built in China. It was a striking scene — one of Silicon Valley's most powerful figures lobbying not for his own products, but for the freedom to use a rival's. It captured, in miniature, the crisis that has crept up on the American technology industry over the past month. The immediate cause of that crisis is a wave of Chinese open-source AI models — software that is free to download, modify and deploy.

One example, Moonshot AI's Kimi K3, is considered powerful enough to compete with proprietary systems from OpenAI and Anthropic in certain applications. Because these models cost users nothing, they undercut the business case for expensive American alternatives. Chip manufacturers such as Nvidia, which profit whenever AI usage grows regardless of whose software runs on the hardware, see opportunity in this. OpenAI and Anthropic, whose revenues depend on subscriptions and API fees, see an existential threat — and have warned that Chinese-made models may carry security risks.

The division runs straight through the White House. Treasury Secretary Scott Bessent has floated the idea of sanctioning Chinese AI firms for alleged intellectual property theft, while Commerce Secretary Howard Lutnick has received letters from startup founders begging him not to block access to open models. Microsoft, Meta, Palantir and Nvidia jointly published a letter urging lawmakers to avoid restricting open-source AI, widening the split further. What is unfolding is less a trade dispute than an identity crisis for an industry that built its political capital on the claim that it alone could outpace China.