Nvidia, Microsoft, Meta, and OpenAI signed the same AI letter to Washington. Anthropic didn't.

A coalition of major American AI companies — Nvidia, Microsoft, Meta, and OpenAI, joined by Mistral and others — signed an open letter to Washington this week warning that any US action against open-weight AI models would weaken American leadership in the field.

The letter, published by Microsoft and signed by 32 organizations total, amounts to a full-throated endorsement of open-weight AI models — the kind any developer can download, run, and customize — at a time when top Trump administration officials have suggested they might take action against Moonshot AI, the creator of Kimi K3, which the letter does not name but which is widely understood to be the implicit target.

In his first-ever post on X, Nvidia CEO Jensen Huang called attention to the coalition's message. "AI will transform every industry, power every company, and be built by every country," Huang wrote. "Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty."

The coalition

The signatories span major US AI labs, cloud providers, and security firms: Nvidia, Microsoft, Meta, OpenAI, Mistral, Palo Alto Networks, Perplexity, and others, according to the letter's full list published by Microsoft.

Two notable absences defined the letter's initial publication: OpenAI and Anthropic. By Friday evening, OpenAI had also signed on, with CEO Sam Altman saying earlier that day he was "glad to see" the support and that he wanted "the US to win in AI both in open source and proprietary models," Altman wrote on X. That leaves Anthropic — maker of the Fable 5 model, which, like OpenAI's GPT-5.6-Sol, is sold via API rather than released as weights — as the only major US frontier lab not to sign.

The policy fight

The letter states: "Our AI leadership will be judged not by one frontier AI model, but whether the United States builds a strong, open ecosystem that diffuses into every sector."

The implicit target is Kimi K3, made by Chinese lab Moonshot AI, whose models top Trump administration officials have suggested could face US action. Kimi K3 is an open-weight model — it can be downloaded and customized by any developer, unlike the proprietary frontier models sold by Anthropic and OpenAI. Both Anthropic and OpenAI would stand to gain if US restrictions impeded Chinese open-weight competitors.

The endorsers from the sidelines

Elon Musk, whose SpaceXAI is not listed as a signatory, voiced support from outside the formal coalition. "This has my full support," Musk wrote on X, quoting Huang's post. "Jensen is right."

Google CEO Sundar Pichai also signaled support without formally signing. "Very happy to support this on behalf of Google," Pichai posted on X. Google is not a formal signatory.

The stakes

The open-weight versus closed-frontier debate is not purely academic. Open-weight models can be deployed sovereignly — a government or company can run them entirely on their own infrastructure — customized for specific industries, and modified without dependency on an API provider. Critics, including those inside the labs that did not sign, argue that publicly released models with few restrictions also lowers the barrier for misuse, since anyone can download a powerful model and adapt it without oversight.

The question the letter poses to Washington is whether the US response to competitive pressure from Chinese open-weight models should be restriction — limiting what American companies and developers can access and build on — or acceleration, ensuring US-built open ecosystems remain competitive globally.