Jensen Huang's First X Post Backs Open Weights. 25 Companies. Here's the Split.
TL;DR
- NVIDIA CEO Jensen Huang posted on X for the first time ever on July 24, 2026. Thirty years of silence, broken with a policy letter, not a product launch. - 25 companies co-signed: Microsoft, Meta, IBM, Dell, Palantir, Hugging Face, Perplexity, and several others. - Huang's quote: "open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty." - The letter has a target.
Thirty years. That's how long Jensen Huang stayed off social media. Then July 24, 2026 happens and the man drops a three-page open-weight policy letter as his opening move on X. Not a GPU announcement. Not a keynote teaser.
A lobbying document.
Wild choice tbh.
The letter, titled "Open Weights and American AI Leadership," came with 25 company signatures attached.
The sell: don't restrict downloadable AI models. The wording they used — "premature restrictions on downloadable AI models". Is doing a lot of heavy lifting.
And then you look at who didn't sign.
That's where this gets real.
---
What Did Jensen Huang Actually Say on X?
Three pages. One core thesis: open-weight models shouldn't get regulated out of existence because they're load-bearing for U.S. competitiveness.
Huang wrote that "AI will transform every industry, power every company. And be built by every country." Big framing. The kind of sentence that sounds like a vision statement until you realize it's also an argument against any country. Including the U.S. — walling things off.
Then the direct quote everyone's been circulating: "Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty."
But here's the thing people keep skipping.
He didn't go full open-source maximalist. The letter explicitly says "The world needs both frontier closed models and frontier open models," which is... a pretty convenient position when your company sells chips to both camps. The policy ask itself warns against "premature restrictions on open models that stifle competition or drive innovation overseas."
Side note: "premature" is doing more political work in that sentence than any single word should.
So the GPU kingpin uses post numero uno to push Washington on open weights. Framed as safety. Framed as sovereignty. And it is those things, sure. But every open-weight release too drives compute demand straight toward NVIDIA's revenue line. More models shipping means more buyers shopping for GPUs. I'm not calling it cynical.
It's just that the interests here overlap in ways worth noticing before you start treating this letter like a manifesto.
---
Who Signed and Who Didn't?
25 signers. Chipmakers. Server vendors. Cloud operators. Security firms. VCs. You get Dell, Microsoft, IBM, and several others.
Now the fun part.
OpenAI, Anthropic, and Google. Nothing. Crickets. Not on the list.
Reports say OpenAI and Anthropic have been actively working lawmakers to clamp down on open-source AI, pointing at risks from strong foreign open models as the reason.
So the companies building closed frontier models want restrictions. The companies selling picks and shovels don't. Shocking, I know.
It's the whole playbook in miniature.
Honestly, just watch what these firms do in Washington.
The X posts are theater. Lobbying: the actual product.
---
Why This Letter Arrived Right Now
Timing's not random.
There are ongoing efforts to restrict or ban certain foreign AI models. The letter doesn't mention specific entities by name. Doesn't need to. Everybody in that room knows exactly which models are being discussed.
Huang himself previously expressed concerns about global AI competition. That puts his position closer to certain framings than to where some other companies have landed on this.
Doesn't make him wrong about open weights necessarily. But it does mean this isn't just a developer-preference debate anymore. Geopolitics is baked in now.
---
What This Means for Your Stack
If open-weight models are running in your production stack, this letter is about whether those weights stay accessible. Next month. Next year. That's the real question.
The distillation fight, the export control chatter, entity list moves, IP enforcement.
Any of it could choke the pipeline. Not abstract. Concrete.
What I'd do right now:
Audit your model dependencies. Which weights are you running, where'd they come from, what breaks if one disappears.
Build fallback plans. Can you swap models without rewriting half your app? If the answer's no, that's a problem.
Track the policy timeline.
The ongoing push for restrictions is active. The 25 signers are pushing back. Your roadmap sits in the middle.
The letter claims open-weight models "enable sovereignty". Governments and orgs running AI on their own infra instead of depending on a single vendor or a foreign provider. Sounds right in theory. Practice is other. If your whole stack runs through one API endpoint, you don't have sovereignty. You've got a subscription with a grace period.
Open weights matter for small operators.
They're the reason a team of five can ship against a billion-dollar compute budget. That access is genuinely worth defending. But defend it with clear eyes about who's standing next to you. Nvidia wants more models as more models means more GPU revenue. Microsoft wants openness everywhere it doesn't compete with closed models. The VCs want cheaper infra for portfolio companies. Your interests and theirs overlap today. Tomorrow is someone else's problem — at least that's how they're treating it.
Read the letter yourself.
Watch Washington over the coming months. Start planning like open weights might not always be a given. Since right now, they might not be.
Sources
- Tom's Hardware: Nvidia and 24 other companies sign open weights letter - Mitrade: Jensen Huang breaks three-decade social media silence - Fortune: Jensen Huang open-source letter
Comments ()