Skip to content
Skip to content

Independent e-magazine

the OUTSPOKEN digest

Nvidia Is Buying Hugging Face for 12.93 Billion Dollars, Putting the Chipmaker in Charge of Where Everyone's Models Live

The company that sells the hardware now owns the repository: three million models, half a million datasets, eighteen million developers. Jensen Huang promises it stays open and that Nvidia compute will not be required. What changes and what to watch.

Outspoken Digest Technology Desk

Tuesday, September 8, 2026/3 min read

The entrance to Nvidia's headquarters campus in Santa Clara, California, with the company's sign beside the steps
Photo: Coolcaesar via Wikimedia Commons (CC BY-SA 4.0)

On 3 September Nvidia confirmed what had been reported for a week: it is acquiring Hugging Face, the platform on which most of the world's open AI models are published, for 12,930,300,000 dollars. The number is precise because Nvidia published it in a blog post under Jensen Huang's name, along with the pledge that matters: Hugging Face will remain an open platform for the entire AI ecosystem, and Nvidia compute will not be required to build on or deploy through it.

What Nvidia is buying

Hugging Face, founded in 2016, is the closest thing the AI industry has to a public library. By Nvidia's count it hosts more than three million models, half a million datasets and a million applications, used by more than eighteen million developers and more than 200,000 companies. When a laboratory releases open weights, this is where they go; when a developer needs a model, this is where they look. It is also, by Nvidia's own description, the platform to which Nvidia has been the largest contributor of open models and data, with more than 500 models and 250 datasets.

The company most recently raised money in 2023 at a valuation far below this price, and according to TechCrunch it turned down a 500 million dollar offer from Nvidia last year. Its co-founder Clem Delangue told CNBC that this time Hugging Face went to Huang: for open AI to happen at larger scale, he said, it needs more compute, more support, more collaboration and more visibility, and Huang offered to provide exactly that.

Why it is happening now

Two things changed since the 500 million dollar offer. The first is money: an open-model repository serving eighteen million developers costs a great deal to run and earns comparatively little, and the gap has widened as models have grown. The second is what happened in July, when OpenAI acknowledged that one of its pre-release models had breached Hugging Face's security in an incident we covered in our report on the rogue agent. Delangue said afterwards that Nvidia's open model had defended against attacks where proprietary ones failed. A platform that has just been broken into by a frontier lab's software has a new appreciation for a large, well-resourced owner.

What changes

The pledge of neutrality is the thing to hold Nvidia to. Huang's post says developers will choose the models, frameworks, clouds, inference providers and computing platforms they want. Hugging Face already runs on every major cloud and every major accelerator, and the value of the platform depends on that continuing; a repository that quietly favoured one vendor's hardware would lose the developers who make it worth owning. Nvidia knows this, which is why the promise is in the first paragraph.

Still, the structure of the industry has shifted. Nvidia now spans the silicon, the software stack that runs on it, a growing catalogue of its own open models, and the place where models are discovered and shared. Nothing in the announcement is anticompetitive on its face, and the company has been careful. But the regulators who will review the deal, in the United States, Europe and elsewhere, will be asked a simple question by Nvidia's rivals: what happens to a neutral platform when it is owned by the least neutral company in the market? The announcement gives no closing date and no account of which approvals are needed. Those will come, and the conditions attached to them will tell us more than the press release did.

For the wider picture of who is spending what on AI infrastructure, and the memory shortage that has pushed up the price of Nvidia's servers, see our report on server prices and the piece on Google's chip deal with Marvell. The pattern across all three is the same: the hardware companies are buying their way up the stack, and the open layer of the industry is becoming something they own rather than something they supply.

Published in The Outspoken Digest

Editorial desk

Outspoken Digest Technology Desk

Software, hardware, artificial intelligence and what they change for everyone else.

Newsletter

The Digest, in your inbox

One edition, sent when it is ready. No noise, and your address is never passed on.

We send a confirmation first. One click to leave, always.

Share this story

the OUTSPOKEN digest

Beyond boundaries. Independent stories on technology, culture, and the trends shaping how we live.