NVIDIA + 24 companies sign open-weights letter opposing AI model restrictions as Jensen Huang joins X

NVIDIA founder and CEO Jensen Huang made his first-ever X post on Friday co-signing a three-page policy letter backed by 25 companies calling on Washington to reject 'premature restrictions on downloadable AI models.' The coalition includes Microsoft, Meta, IBM, Dell, Palantir, Hugging Face, and AI labs Mistral, Black Forest Labs, and Arcee AI—but notably excludes OpenAI, Anthropic, and Google, signaling a divided tech industry on open-weight policy.

The letter comes four days after reports that the Trump administration was reviving a push to ban Chinese AI models, specifically citing concerns about models like DeepSeek. The signatories argue that open models strengthen safety, cybersecurity, innovation diffusion, and national sovereignty—a direct rebuttal to arguments that downloadable models pose national security risks. Huang tweeted that 'one in every four tokens generated today comes from an open model,' reframing open-source AI as already dominant infrastructure.

The letter also calls for expanded compute access for startups and researchers, public investment in shared datasets, and frameworks that don't treat model distillation (training one model on another's outputs) as misappropriation. Treasury Secretary Scott Bessent had claimed days earlier that U.S. officials found watermarks from American LLMs in Chinese systems, but Huang told Axios that American firms should be allowed to use Chinese models and that claims of backdoors were misconceptions.

The absence of OpenAI, Anthropic, and Google—who prioritize closed model deployment—suggests a strategic rift in the AI industry. For practitioners, the letter's policy asks frame open weights as essential infrastructure that should remain free from supply-chain weaponization, even as hyperscalers race to lock customers into proprietary inference APIs. The fight over downloadable models is now explicitly political.