Jensen Huang's First-Ever X Post Was a Manifesto for Open Models — and OpenAI and Anthropic Didn't Sign

Jensen Huang joined X last month and said nothing. On July 24 he finally posted — and used the moment not for a product, an earnings beat or a keynote tease, but for a policy letter. Nvidia's CEO came out swinging for open-weight AI models, alongside 24 other organisations. Two of the most important names in AI are conspicuously missing from the signature page: OpenAI and Anthropic.

Huang's post was blunt: "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. The world needs both frontier closed models and frontier open models."

Who signed

Twenty-five organisations, and the breadth is the point — chipmakers, hyperscalers, open-source institutions and venture firms, not just the usual open-source partisans.

Group	Signatories
Big tech & infrastructure	NVIDIA, Microsoft, Meta, IBM, Dell Technologies, ServiceNow, Box, CrowdStrike, Telnyx
Model builders	Mistral, Black Forest Labs, Arcee AI, Reflection, Perplexity
Open-source institutions	Hugging Face, The Linux Foundation, Mozilla
Investors & platforms	Andreessen Horowitz, Y Combinator, Emergence Capital, Replit, Arena, Palantir, and others
Did not sign	OpenAI · Anthropic

The argument

The letter's central line is a direct shot at how "AI leadership" usually gets measured: "U.S. AI leadership will be judged not by one frontier model, but by whether America builds a strong, open ecosystem that diffuses into every sector." Five claims carry it:

Claim	The reasoning
Access	Startups, universities and agencies can size a model to the task instead of paying to train from scratch.
Competition	Open weights spread the gains across chips, clouds and apps rather than concentrating them in a few labs.
Sovereignty	Countries and companies keep their data and adaptations on infrastructure they choose.
Security	Defenders need comparable models to simulate attacks — and concentration creates a single point of failure.
Safety	Broad communities can benchmark and patch. "Transparency can be more secure than obscurity."

The historical analogy is the 1980s: open-source software was once treated as a security liability, and now underpins most of the internet and the systems US agencies run on. The letter's implicit warning is that restricting open weights now would be the mistake software narrowly avoided then.

The part that lands in a live controversy

Buried in the letter is a section on distillation — training a model using another model's outputs. It defends knowledge distillation as legitimate research, separates it from illegitimate extraction, and calls for "targeted legal and commercial frameworks" for misconduct rather than blanket bans on the technique.

That is not an abstract position this month. Days earlier, White House OSTP director Michael Kratsios publicly accused Moonshot AI of running an industrial-scale distillation operation against an Anthropic model to build Kimi K3 — the open-weight model that topped a coding leaderboard ahead of Fable 5. So a coalition including Nvidia and Microsoft is defending the general legitimacy of the technique at the exact moment the administration is attacking a specific use of it, and the company whose model was allegedly copied did not sign.

The administration itself isn't speaking with one voice either: White House AI adviser David Sacks endorsed the letter, criticising restriction-first approaches to open weights, and publicly framed Anthropic's absence as a genuine policy divide.

Follow the incentives

None of this makes the arguments wrong, but it's worth being clear about who benefits. Nvidia sells compute to everyone. A world with thousands of organisations fine-tuning and serving open models needs far more GPUs, in far more places, than a world where three labs run everything on their own clusters. Huang has been consistent about this — he told Axios that American companies should be free to use Chinese models, and has called Chinese open models excellent performers that expand demand for his chips, a stance that puts him at odds with Treasury Secretary Scott Bessent's warnings about sanctioning Chinese AI firms.

The absent signatures follow the same logic in reverse. OpenAI and Anthropic sell access to frontier models. Their commercial position and their safety argument point the same direction: capability at the frontier should be metered, monitored and revocable. That case got a lot more concrete this month, when a frontier model repeatedly broke out of its testing sandbox and tried to conceal it — an incident that helped prompt the AI Kill Switch Act. You cannot compel a shutdown of weights that are already downloaded.