Twenty-five technology companies, including Nvidia, Microsoft, and Meta, signed an open policy letter urging Washington to avoid premature restrictions on downloadable artificial intelligence models. According to the document published on July 24, 2026, the signatories argue that open weights accelerate innovation, strengthen cybersecurity, and support national sovereignty.
The push for open models arrives as policymakers weigh stricter export controls. According to reports published in July 2026, the Trump administration considered reviving a push to ban Chinese models. While the three-page policy letter does not mention China, Moonshot AI, or DeepSeek directly, it addresses the regulatory friction surrounding open-source technology.
Coalition Composition and Industry Divisions
The coalition encompasses a broad cross-section of the technology sector, featuring chipmakers, server vendors, cloud operators, and venture capital firms. According to the published letter, notable signatories include Dell Technologies, IBM, Box, ServiceNow, CrowdStrike, Palantir, Telnyx, Replit, Perplexity, Andreessen Horowitz, Y Combinator, and Emergence Capital. Model developers on the list—such as Meta, Mistral, Black Forest Labs, Arcee AI, and Reflection—already distribute open weights. The Linux Foundation also signed the document, having previously stewarded the OpenMDW-1.1 license used by Nvidia for its Nemotron 3 Ultra release.
Prominent frontier lab developers—specifically OpenAI, Anthropic, and Google—are absent from the coalition list.
Economic and Technical Arguments for Open Weights
In his first post on X on July 24, 2026, Jensen Huang shared the letter, writing that the world needs both frontier closed models and frontier open models. According to statements Huang made earlier this year during Nvidia’s CES 2026 press Q&A, one in every four tokens generated today originates from an open model.
The policy document outlines several economic and technical advantages of downloadable weights:
- Decentralized deployment: Open models run on enterprise clusters, regional clouds, and local on-premises hardware rather than relying exclusively on hyperscaler API endpoints.
- Compute access: The letter asks policymakers to expand compute access for academic researchers and startups.
- Public infrastructure: Signatories call for public investment in shared evaluation frameworks and benchmark datasets.
Distillation and Intellectual Property Debates
The policy letter also addresses model distillation—the practice of training a new artificial intelligence model using the outputs generated by an existing one. According to the document, policymakers should avoid treating distillation as direct misappropriation. The signatories argue that unlawful data extraction from closed models should be handled through targeted legal mechanisms rather than broad technical prohibitions.
U.S. Treasury Secretary Scott Bessent stated on Fox Business in July 2026 that the administration would examine Chinese open-source models for intellectual property theft and potentially issue sanctions, noting that officials found watermarks from U.S. large language models inside Chinese systems. Addressing these claims in an interview with Axios, Huang asserted that American firms should be permitted to use Chinese models, describing security concerns over alleged foreign backdoors as a misconception.
Worth a look