Microsoft and Nvidia lead open-weight AI letter to US

This letter is the clearest signal yet that big tech wants to lock in political space for open-weight models before Washington reacts to the Hugging Face incident and Chinese open models with sweeping controls. When Microsoft, Nvidia, Meta, IBM, Palantir, Hugging Face, Mistral and Cohere all sign the same document, it is not just PR, it is coordinated industrial policy. They are arguing that American AI leadership should be measured by the breadth of an open ecosystem, not just by who trains the single largest frontier model, and they are explicitly asking Congress not to collapse open weights into the same risk bucket as closed, frontier systems.

For the race to AGI, the stakes are straightforward. If open-weight models remain legal and well supported, we move toward a world where highly capable systems are widely replicated, audited and adapted by hundreds of actors, not hoarded by a few labs. That could accelerate capability diffusion, including agentic systems built on top of Nemotron, Gemma, GLM or Kimi-class models, and it gives enterprises and governments a credible alternative to renting everything from closed labs. At the same time, it complicates safety governance, because policing behavior across a sprawling open ecosystem is much harder than regulating a handful of centrally hosted frontier models.

Strategically, this letter also signals a growing split between GPU vendors and cloud hyperscalers that profit from open ecosystems and frontier labs whose business model depends on scarcity. If lawmakers side with the signatories, we will likely see more capital flow into open-weight stacks and sovereign AI projects, especially outside the US, making the AGI race more multipolar.