Nvidia open source AI push draws Meta, Microsoft into Washington fight

Nvidia open source AI policy moved into the spotlight Friday as 25 companies, including Meta, Microsoft, Andreessen Horowitz, Hugging Face, IBM and Dell, published a joint letter urging Washington not to restrict open-weight models. For retail investors watching the AI trade, the message is straightforward: some of the biggest companies in the sector think open access is part of America’s competitive edge, not a risk to be shut down.

The letter, titled “Open Weights and American AI Leadership”, argues that limits on open models would not make the U.S. safer. Instead, the signers say restrictions could concentrate the AI market around a smaller group of closed-model companies.

OpenAI and Anthropic were not listed among the signatories.

What are open-weight AI models?

Open-weight models are AI systems whose model weights, the numerical settings that help determine how the system responds, are made publicly available. That lets developers download the model, run it on their own infrastructure and modify it, instead of accessing it only through a company-controlled app or paid application programming interface, known as an API.

The letter compares the current AI debate to the open-source software movement of the 1980s. The companies argue that U.S. AI leadership should be measured by whether the country builds a broad ecosystem that spreads across industries, rather than by whether one frontier model sits at the top.

“Our AI leadership will be judged not by one frontier AI model, but by whether the United States builds a strong, open ecosystem that diffuses into every sector,” the letter says.

Why Nvidia, Meta and Microsoft are making the case now

Nvidia CEO Jensen Huang shared the letter in his first post on X. Huang wrote that AI will reshape industries, support companies and be built by countries around the world. He also said open models improve safety and cybersecurity, speed innovation and help enable sovereignty, meaning the ability for governments or organizations to run key technology on their own terms.

Microsoft CEO Satya Nadella also posted the letter on X. Nadella described open models as essential to a healthy AI ecosystem while also saying national security must be protected.

That framing matters because AI is now a policy issue as much as a product race. If regulators make it harder to release or use open-weight systems, the business impact could stretch across chipmakers, cloud providers, model developers and startups building on top of AI tools.

The signers’ core argument is that open models create more competition and give more organizations a way to build AI products without depending entirely on closed systems. Closed models, such as OpenAI’s GPT and Anthropic’s Claude, are accessed through the companies that operate them rather than downloaded and changed directly by outside users.

The letter does not ask Washington to ignore safety concerns. Its argument is that openness can support security by letting more researchers, companies and governments test, adapt and improve models. The companies also say open systems can help American technology spread more widely across sectors.

For investors, the policy fight is another reminder that AI leadership is not only about who has the most powerful model. It also depends on distribution, developer adoption, infrastructure and regulation, all of which can shape which companies capture value as AI spending keeps moving through the market.

This story draws on original reporting from Decrypt.