The Trump administration considered aggressive sanctions, trade blacklists, and cloud-service bans targeting Chinese open-source AI models in recent weeks, before backing off after a coordinated pushback from Nvidia, Meta, Microsoft, and Google, according to New York Times reporting.
The internal White House debate, led by Chief of Staff Susie Wiles, Treasury Secretary Scott Bessent, and Commerce Secretary Howard Lutnick, weighed options including sanctions on Chinese makers of open-weight models and a ban on US cloud providers doing business with those firms. Five sources cited by the Times said the administration also discussed restricting which models federal agencies could use and whether Commerce should impose new export rules.
The confrontation was triggered by the rapid rise of Chinese open models, particularly Moonshot AI's Kimi K3, which now matches top US models across several benchmarks. Michael Kratsios, director of the White House Office of Science and Technology Policy, publicly accused Moonshot of distilling Anthropic's Fable model. Bessent floated sanctions the same day.
The industry response split along predictable lines. OpenAI and Anthropic, both of which keep their models closed, pressed for restrictions by citing national security. On the other side, Nvidia, Microsoft, Google, and Meta argued that open models accelerate innovation and cyber defense and warned that bans would lock in the market position of the two IPO-bound AI labs.
Nvidia CEO Jensen Huang became the most visible counterweight. On July 24, he posted on X for the first time to defend open models. Nvidia also expanded an alliance for open security tools from a handful of members to more than 230.
Christopher Padilla, a former Commerce Department official, told the NYT the dispute was a "cage fight over the future of the AI industry," adding that it was mostly about money and market power with the security angle tacked on.
After the industry outcry, the administration shifted focus to making American models more competitive rather than restricting Chinese ones. Industry officials told the Times they do not expect any hard measures before Chinese President Xi Jinping's scheduled visit in September.
The episode illustrates a widening rift in Silicon Valley over open-weight AI. OpenAI and Anthropic have staked out positions that open models pose unacceptable security risks, while the infrastructure and platform giants whose businesses depend on broad AI adoption see openness as essential.
Sources
