Huang's open-weights letter doubles to 50, Anthropic still out

A one-day doubling of a policy letter's signatory list is unusual enough to be worth pausing on, especially when the letter is about who gets to decide which AI models Americans can download. Jensen Huang's 'Open Weights and American AI Leadership' letter went out on July 24 with 25 companies attached, and by the next day Forbes reported it had reached 50, with OpenAI, Google, AMD, Cisco, Cloudflare, GitHub, Block and Ollama among the new joiners.

The coalition's framing is that Washington should not impose sweeping restrictions on models whose trained parameters are made publicly available, arguing that open weights accelerate innovation, strengthen cybersecurity and support national sovereignty. CNBC tied that ask directly to policymakers weighing action against open-weight models, and Huang's post, reportedly the first of his life on X, cleared 11 million views and turned what would have been a routine industry letter into something closer to a Silicon Valley alignment vote.

The two names still not on the letter are what makes it interesting. Amazon and Anthropic did not sign in the initial 25 and did not sign in the follow-on tranche either. Anthropic has, per the reporting, taken an increasingly hard line against open-weight models, and the company is reportedly preparing for a massive initial public offering, so its absence reads more like a stance than an oversight. For a lab whose entire brand is AI safety, being the isolated closed-model holdout after even OpenAI signed is a genuine posture, not a coincidence.

The honest caveat is that this is a lobbying document, not a law. What the reporting does not spell out is exactly which policy proposal the letter is responding to. Tom's Hardware frames it against Washington weighing a Chinese AI model ban, there is a separate industry fight over a 'distillation clause,' and none of the coverage yet explains why OpenAI reversed from initial absence to a signature within roughly 24 hours.

What is worth watching is whether policymakers read 50 companies on one side and two on the other as industry consensus, or whether Anthropic's safety-first framing plus Amazon's weight in cloud is enough to keep the closed-model position credible in Washington. The coalition has locked in the numbers. The regulatory argument over what 'open' actually means, and what a distillation clause would allow, is where the next round is going to happen.