Open-Weight AI Models Backed by Nvidia and Microsoft

Open-weight AI models received public support on July 24, 2026, when about 25 technology companies led by Nvidia (NVDA) and Microsoft (MSFT) issued an open letter urging U.S. policymakers to avoid premature or sweeping restrictions. The letter could influence regulation and competitive access in the AI sector.

Industry Coalition Letter and Policy Requests

The coalition’s letter, known as Open Weights and American AI Leadership, included signatories such as Meta, Palantir, IBM, Andreessen Horowitz, Hugging Face, Mozilla, the Linux Foundation, and Dell. Notably, OpenAI and Anthropic did not join the letter.

The group defined open-weight models as AI systems whose underlying parameters, or weights, can be downloaded, inspected, modified, and run on local infrastructure. The letter urged lawmakers to reject early restrictions on these models and oppose broad limits on distillation techniques, which involve training or fine-tuning models using proprietary data. Instead, the coalition recommended addressing unlawful uses through targeted legal and commercial frameworks.

The signatories argued that open weights promote competition and broader access to AI benefits. They called on Washington to expand public computing resources and shared training infrastructure to help smaller developers remain competitive. Nvidia Chief Executive Jensen Huang shared the letter in his first post on X, and Microsoft Chief Executive Satya Nadella amplified the appeal on social media.

U.S. Policy and Security Developments

On June 2, 2026, President Trump signed Executive Order 14409, “Promoting Advanced Artificial Intelligence Innovation and Security,” establishing a voluntary 30-day cybersecurity review for “covered frontier models.” Under this framework, developers can grant the government access before broad release. The order explicitly stops short of mandatory licensing or preclearance and directs Treasury, the National Security Agency (NSA), and the Cybersecurity and Infrastructure Security Agency (CISA) to help form an AI cybersecurity clearinghouse.

On June 12, 2026, the Commerce Department issued an export-control order requiring Anthropic to take its Fable 5 and Mythos 5 models fully offline, demonstrating federal willingness to intervene directly in frontier-model deployment.

Federal officials have discussed a range of tools—including procurement restrictions, Entity List additions, sanctions, security warnings, and liability rules—to discourage U.S. hosting or use of advanced foreign open-source models. The White House has publicly denied drafting a specific executive order targeting open-source AI.

Treasury Secretary Scott Bessent said on July 21 that the U.S. would examine Chinese open-source AI models for intellectual-property theft and could sanction companies found to have stolen American IP. Chinese open-weight models such as Moonshot AI’s Kimi K3, released in July 2026, have narrowed the performance gap with leading U.S. systems and intensified regulatory concern.

Developers focused on closed frontier systems, including OpenAI and Anthropic, have supported tighter controls and cooperated with government access regimes. OpenAI agreed on June 25 to stagger the rollout of GPT-5.6, initially offering it only to government-approved customers.

The coalition’s call for a plural frontier—maintaining both open and closed frontier models—and its emphasis on precision regulation targeting verifiable risks, high-risk deployments, IP theft, and misconduct present an alternative to origin-based curbs. The path Washington chooses could shape U.S. AI policy and cloud-compute demand for major providers.