EditNVIDIA-led coalition, backed by European firms, urges U.S. policymakers to embrace open-weight AI
France’s Mistral and Germany’s Black Forest Labs join U.S. giants in arguing that open models strengthen leadership, safety, and sovereignty
On 24 July NVIDIA CEO Jensen Huang used his first-ever post on X to share a joint letter titled “Open Weights and American AI Leadership.” The document, signed by more than two dozen organizations spanning chips, models, infrastructure, and venture capital, argues that downloadable, inspectable, and modifiable AI models are essential for sustained U.S. technological edge, economic diffusion, competition, and even cybersecurity.
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.… pic.twitter.com/t02bi51N4C
— Jensen Huang (@JensenHuang) July 24, 2026
Elon Musk quickly endorsed it with a clear statement: “This has my full support. Jensen is right.” Microsoft CEO Satya Nadella also amplified the message.
Notably, the coalition is not purely American. European companies including France’s Mistral (one of Europe’s leading AI labs and a consistent publisher of open-weight models) and Germany’s Black Forest Labs appear among the signatories alongside NVIDIA, Microsoft, Meta, IBM, Palantir, Hugging Face, the Linux Foundation, a16z, Y Combinator, and others. Their participation signals a broader Western alignment on the value of open ecosystems at a moment when questions of technological sovereignty are rising on both sides of the Atlantic.
The letter draws a deliberate parallel to the 1980s open-source software movement. It contends that open-weight models expand access for startups, enterprises, universities, and public institutions; foster competition that prevents concentration of power; give organizations greater control over their data and deployments; and improve safety through community scrutiny rather than relying solely on closed systems that create single points of failure. Policymakers, it says, should expand compute access, invest in shared datasets and tools, and avoid “premature restrictions” on open models or legitimate techniques such as distillation.
The timing is pointed. The letter arrived days after China’s Moonshot AI released Kimi K3, an open-weight model claiming near-frontier performance. Washington has been weighing potential measures related to Chinese models, heightening industry concerns about overregulation that could hinder domestic innovation or push activity overseas.
The signatory list also highlights an industry fault line. While many open-oriented or infrastructure players signed promptly, leading closed-frontier labs Anthropic and (initially) OpenAI and Google were absent. OpenAI later signaled support or joined in some reports, with CEO Sam Altman stating he wants the U.S. to succeed in both open and proprietary models. Anthropic has remained outside the coalition.
For European participants such as Mistral, the stance aligns with long-standing arguments for strategic autonomy: the ability to run, inspect, and adapt models without dependence on a small number of foreign closed providers. By backing a letter framed around American leadership, these firms underscore that open weights can serve complementary goals of U.S. competitiveness and European technological independence.
Whether the letter influences actual policy remains to be seen. What is already clear is that a significant cross-section of the AI stack – spanning Silicon Valley hardware leaders, American platforms, and key European model developers – has publicly drawn a line against sweeping limits on open-weight systems. The debate over openness versus control is no longer abstract; it now has a high-profile, multi-continental coalition attached to it.
Caption: Nvidia CEO Jensen Huang speaks during the Korea AI Ecosystem Reception in Seoul, South Korea, 8 June 2026. EPA/YONHAP
Updated: July 25, 2026 - 14:51
