Nvidia, Microsoft, and Meta support OpenAI, but OpenAI did not reciprocate.

      The AI industry has officially taken a stance. On Friday, 25 companies and organizations released an open letter titled “Open Weights and American AI Leadership,” calling on Washington not to impose restrictions on open-weight AI, according to Business Insider.

      The list of endorsers is notable, featuring major names like Nvidia, Microsoft, Meta, Mistral, Palantir, IBM, Andreessen Horowitz, Hugging Face, Mozilla, and the Linux Foundation. Nvidia's CEO Jensen Huang introduced the letter with his inaugural post on X, which attracted over 11 million views. “For my first post, I’m sharing a letter Nvidia signed on why open models are important,” he stated. “The world requires both frontier closed models and frontier open models.”

      The letter traces its roots back to the 1980s, presenting open weights as a natural successor to the open-source software that now “underpins most of the internet.” It argues that America's leadership in AI should not be measured by a single frontier AI model, but rather by the establishment of a robust, open ecosystem.

      Notably absent from the signers are OpenAI, Anthropic, and Google, which are leaders in closed models. Each of these companies would benefit if Washington were to block Chinese open-model developers like Moonshot, whose Kimi K3 is recognized as the world's most potent open-weight model. Essentially, the letter represents the broader industry positioning itself against them, coming just a day after nearly 200 startups voiced similar concerns to the White House.

      The proponents of closed models did not oppose this initiative. Altman expressed his satisfaction with the support, while Microsoft's Satya Nadella endorsed the message on the same day, as reported by CNBC. Even Elon Musk, whose SpaceX did not sign the letter, showed support by quote-tweeting Huang: “This has my full support. Jensen is right.”

      The letter makes a striking assertion by directing the safety argument towards closed labs. It claims that relying solely on closed models “is not inherently safe,” as they “can be breached, misused, or fail in ways that outsiders cannot detect.” This concentration of AI within a few closed models creates “single points of failure.”

      In contrast, they argue, openness “may be one of the most important paths to AI safety.” This directly counters the stance of OpenAI and Anthropic, who have emphasized security concerns as justification for limiting Chinese open models.

      The timing of the letter is also significant. This week, Michael Kratsios from the White House accused Moonshot of developing the Kimi K3 by extracting insights from Anthropic’s Fable 5. Treasury Secretary Scott Bessent suggested imposing sanctions for such distillation, which he equated with theft.

      While the letter acknowledges distillation as “a widely used technique,” it does not reject all potential abuses. It advises policymakers not to “conflate legitimate model-development techniques with misappropriation,” adding that unlawful extraction requires “targeted” regulations rather than broad restrictions.

      There are self-serving interests at play, along with a challenging reality. Nvidia provides chips to a variety of clients, while Meta, Mistral, and Hugging Face develop open models. All these companies stand to benefit from the proliferation of open AI. Moreover, the US has fallen behind in the very area the letter promotes, with China now taking the lead in a sector that America once pioneered.

      The signatories urge Washington to maintain a “frontier plural” approach and refrain from “premature restrictions” that could “stifle competition or drive innovation overseas.” The letter raises an unresolved question: Can America still triumph in a race it is currently trailing?