Fifty technology companies — including Nvidia, Microsoft, Meta, OpenAI, and Google — have signed a joint letter asking Washington not to restrict open-weight AI models. Amazon and Anthropic are absent from every version of the letter; neither has publicly explained why. The vendor split is concrete intelligence for any operator whose AI strategy depends on open-weight models or whose primary AI vendor is Anthropic.

Key takeaways:

• On July 24, Nvidia CEO Jensen Huang used his first-ever X post to publish “Open Weights and American AI Leadership,” co-signed by 25 companies and hosted on the microsoft.com letter page.
• Within 24 hours the list grew to 50; OpenAI and Google joined the second wave alongside AMD, Cisco, Cloudflare, GitHub, Block, and Ollama — per Forbes’s July 25 tracking.
• Amazon and Anthropic are absent from every version, with no public statement from either on this specific letter.

Why Did This Letter Appear Now?

The timing is direct. Trump administration officials have publicly suggested action against Moonshot AI, the Chinese company behind Kimi K3 — whose 2.8-trillion-parameter weights are scheduled for public release July 27; the Kimi K3 API has been live since July 16, but the full weights have not yet been published. Earlier this month, US Treasury Secretary Bessent warned the US could sanction Chinese AI companies over IP theft allegations and asked whether businesses using Chinese AI should disclose it.

The letter argues against “premature restrictions” on open-weight AI broadly, without naming China, Moonshot, or DeepSeek. Any Washington action restricting Chinese open-weight model access would benefit vendors who sell closed frontier access instead. The letter grew from 25 to 50 signatories in one day.

What Does Amazon and Anthropic’s Absence Mean?

Neither company has commented on the letter. Forbes’s July 25 analysis identifies two readings that may both be true: a commercial-interest reading (Anthropic sells closed frontier model access; Amazon has committed up to $25 billion in Anthropic and is the company’s primary compute provider) and a principled safety reading (Forbes notes Anthropic has argued publicly that model weights, once released, cannot be recalled). Absence is not a statement. It does not establish that either company opposes open weights or supports a specific restriction.

Operators do not need to assign a motive. They do need to know where their vendors sit on a policy question that could reshape their future model options.

What to do now: Map your current AI model dependencies — which workflows and pipelines rely on open-weight models, and which rely on vendors not represented in this coalition. No restriction is in force today. Before committing production workloads to any model that could come under US policy review, ask your legal and compliance teams what your third-party AI risk policies require. For operators with Anthropic or AWS contracts, raise open-weight model portability and continuity at your next vendor review: what happens to your workflows if a model you rely on is restricted by government action?

Watch for: a formal Washington proposal or executive order responding to the letter. That is the signal that turns a policy debate into an operator decision.

---

FAQ

What is the “Open Weights and American AI Leadership” letter? A joint statement by 50 US technology companies — including Nvidia, Microsoft, Meta, OpenAI, and Google — urging policymakers not to impose “premature restrictions” on downloadable open-weight AI models. Nvidia CEO Jensen Huang published it July 24 as his first X post; it is hosted at microsoft.com/en-us/corporate-responsibility/topics/open-weight/.

Does this letter change anything for operators today? No restriction is in place. But the letter signals Washington is actively debating open-weight AI access — with Kimi K3’s July 27 scheduled weight release as a focal point — and that your AI vendor’s position on this debate is now a legitimate procurement diligence question.