Leading technology companies have called on U.S. policymakers to protect access to open-weight artificial intelligence models rather than impose sweeping restrictions. The appeal comes as the Trump administration considers measures targeting advanced Chinese systems that have matched or exceeded some American counterparts at lower cost.

Nvidia CEO Jensen Huang used his first post on the platform X to share an open letter signed by Nvidia, Meta, Microsoft, Dell, Palantir, IBM, Hugging Face, Andreessen Horowitz and roughly 20 other organizations. The document, focused on open weights and American AI leadership, argues that these models form a critical foundation for U.S. competitiveness.

What Open-Weight Models Are and Why Companies Defend Them

Open-weight models release the underlying numerical parameters that determine how an AI system responds. Users can download them, run them on their own infrastructure, customize them and inspect their behavior. This differs from closed models offered only through proprietary applications or paid interfaces.

The letter compares the current moment to the rise of open-source software in the 1980s. Signatories contend that U.S. leadership will be measured not by any single frontier system but by whether the country builds a broad, open ecosystem that reaches every industry. They state that open models strengthen safety and cybersecurity by allowing many independent teams to examine code and identify flaws, accelerate innovation and technology diffusion, and support national sovereignty by giving organizations greater control over data and deployment.

Microsoft has advanced a parallel case, saying American leadership depends on an open, resilient ecosystem rather than one dominant model. Venture investors including Bill Gurley and Ben Horowitz have echoed the view that open models provide a competitive edge and prevent benefits from concentrating among a small number of firms.

Policy Debate Intensifies After Chinese Model Releases

The letter arrives as administration officials examine possible restrictions on Chinese open-weight systems. Moonshot AI’s Kimi K3, a model with 2.8 trillion parameters released earlier in July, has ranked highly on certain benchmarks and is offered at substantially lower prices than leading U.S. alternatives. Officials have raised concerns about how the system was developed.

White House Office of Science and Technology Policy Director Michael Kratsios has alleged that Moonshot built Kimi K3 by distilling Anthropic’s Fable model at industrial scale and obtained servers fitted with export-restricted Nvidia chips through third countries. Treasury Secretary Scott Bessent has stated that companies found to have stolen U.S. intellectual property could face sanctions and Entity List designations, while distinguishing lawful open-source practices from unauthorized extraction of American technology.

The coalition of companies draws a clear distinction on distillation, the practice of using one model’s outputs to train or improve another. They describe it as a longstanding and legitimate technique for model improvement, evaluation and validation that has driven progress since the open-source software era. They urge policymakers to address proven unlawful conduct through targeted legal and commercial tools rather than broad prohibitions on techniques essential to innovation.

Industry Split and Startup Concerns

OpenAI and Anthropic did not sign the letter. Those firms have warned that strong Chinese open models present risks. The absence of the two major closed-model developers underscores a divide within the industry over the degree of openness that should characterize advanced AI.

Smaller companies have also weighed in. The Little Tech Association, representing nearly 200 startups including Y Combinator and Proton, sent letters to the White House and Commerce Department arguing that cutting American developers off from foreign models would stifle competition, entrench incumbents and function as a tax on intelligence. Startup founders have said restrictions could force many firms to rely exclusively on a handful of large providers.

Elon Musk publicly backed the letter, stating it has his full support and that Huang is right. The discussion continues against the backdrop of an earlier U.S. government AI action plan that had described open models as a strategic asset.

Signatories acknowledge that once model weights are released they cannot be recalled and that real risks exist. Their preferred response is focused safeguards against demonstrated harms rather than categorical bans. The coming weeks will test whether policymakers adopt that narrower approach or move toward wider limits on open-weight systems.