Jensen Huang sat on the sidelines of X for years, and on Friday the Nvidia CEO finally posted for the first time. He didn’t use it to talk or sell GPUs. Instead he shared a policy letter making the case that open-source AI models are key to keeping the U.S. ahead in the AI race.
“For my first post, I’m sharing a letter @NVIDIA signed on why open models matter,” Huang wrote. “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.” His post from 6am PT has already seen over 14 million views on X.
The letter is called “Open Weights and American AI Leadership” and it has 25 signatories. Along with Nvidia you’ve got Microsoft, Meta, IBM, Dell Technologies, Palantir, Mistral, Mozilla, Hugging Face, Perplexity, CrowdStrike, The Linux Foundation, Andreessen Horowitz, and Y Combinator. Who’s not on it? OpenAI, Anthropic, Google, and SpaceXAI (formerly xAI), the four labs currently racing each other hardest on frontier (top-tier basically) models.
So what’s an open-weight model? It’s one anyone can download, poke at, tweak, and run on their own hardware. The companies behind the letter say that’s what lets startups, universities, hospitals, and small businesses actually use advanced AI without training their own model or paying top dollar every time they run a task. They compare it to the open-source software movement of the 1980s, which ended up quietly running much of the internet and a lot of US government systems.
On security, the letter flips the usual argument. Instead of treating open weights as something dangerous to lock down, the group says stuffing all the advanced AI behind a handful of closed models is the real risk, since it creates a few big single points of failure nobody outside those companies can test. Their view is that a whole community banging on a model finds bugs faster than one closed team ever will. There are downsides–once weights are out there, the developer loses control and modified versions are tough to track.
The group wants more compute for startups and researchers, money for shared datasets and testing tools, and no rushed restrictions that push open model work to other countries (which in practice means China).
What does this mean for Canada? Open weights would let Canadian companies download and run advanced AI on their own servers, which fits right into the whole sovereign AI push to keep data on home soil. It also makes Ottawa’s $2 billion sovereign compute spending worth something, since a Canadian supercomputer needs capable models to actually run. The catch is that Canada has no say in any of this, so if Washington clamps down on open weights, Canadian startups lose their cheapest route to advanced AI.
Want to see more of our stories on Google?
P.S. Want to keep this site truly independent? Support us by buying us a beer, treating us to a coffee, or shopping through Amazon here. Links in this post are affiliate links, so we earn a tiny commission at no charge to you. Thanks for supporting independent Canadian media!
