Nvidia Is Buying Hugging Face for $12.9 Billion. Now It’s Betting the Open-Model Ecosystem Stays Loyal.
Nvidia is buying Hugging Face, the platform that has become the default hosting and distribution point for open AI models and datasets, in a deal worth $12,930,300,000. Nvidia disclosed the total in a Form 8-K filed with the SEC on September 2, 2026, and both companies confirmed the acquisition publicly the following day.
The structure combines roughly $11.9 billion in cash and stock paid to Hugging Face’s stockholders with a separate equity-based retention pool of up to $1,000,000,000 for employees who join Nvidia — the exact split behind the “$12.9 billion” figure TechCrunch reported and the “$13 billion” figure Bloomberg used, both rounding the same SEC filing.
It is a full acquisition, not the minority stake Nvidia has held since 2023, and it is expected to close in the first half of 2027 — “subject to the satisfaction or waiver of customary closing conditions, including receipt of required regulatory approvals,” per the filing. That approval has not yet been granted by anyone.
- $12,930,300,000 — total deal value Nvidia disclosed in its Sept. 2, 2026 SEC Form 8-K — reconciling the “$12.9 billion” (TechCrunch) and “$13 billion” (Bloomberg) headline figures
- 18M+ — developers on Hugging Face's platform, which also hosts 3 million-plus models and 500,000-plus datasets — Nvidia
- $235,000,000 — Hugging Face's 2023 Series D round, where Nvidia first bought in as a minority investor
- ~4% — NVDA's share-price jump, to roughly $225.96, after The Information's Aug. 27, 2026 report on the deal
- First half of 2027 — Nvidia's expected close date — “subject to … required regulatory approvals”, none of which has been granted yet — SEC 8-K
Nvidia already knew Hugging Face’s business well before this deal. It first invested in 2023, when Hugging Face raised $235,000,000 in a Series D round at a $4.5 billion valuation, buying in as one of several strategic backers rather than a controlling investor. This acquisition replaces that toehold with full ownership: under the 8-K’s terms, Hugging Face stockholders receive the bulk of the $12,930,300,000 in cash and stock, while a separate pool of up to $1,000,000,000 in Nvidia equity is reserved to retain the employees who join the company after close. Buying the company and then separately paying to keep its people is a standard structure for acquisitions at this scale — it is not standard to see it applied to a deal this size in AI infrastructure.
Nvidia’s own filing language is careful on timing: the deal is expected to close in the first half of 2027, “subject to the satisfaction or waiver of customary closing conditions, including receipt of required regulatory approvals.” That single clause is doing real work. A transaction of this size mandatorily triggers a Hart-Scott-Rodino antitrust filing in the United States, and given Nvidia’s position in AI computing, a European Commission merger review is widely expected as well — neither has happened yet, and neither regulator has said anything publicly about this specific deal as of this writing. Section 04 below covers what that review is expected to look at.
The strategic logic starts with scale. Hugging Face hosts more than 18 million developers, over 3 million models, more than 500,000 datasets, roughly 1 million applications, and more than 200,000 companies that use the platform, according to Nvidia’s own announcement. Nvidia has spent a decade building the hardware nearly every AI lab trains and runs on; this deal moves it into owning the largest distribution layer those same labs use to publish and find models. That is vertical integration in the most literal sense — from the chips underneath the industry to the shelf where its output gets displayed.
There is also a defensive angle. OpenAI, Google, Amazon, and Anthropic are all building or expanding custom AI chips specifically to reduce how dependent they are on Nvidia hardware. Owning a thriving open-model ecosystem keeps developers working in Nvidia’s orbit even on a platform that is, by design, hardware-agnostic. Fortune has framed the deal as a way for Nvidia to redeploy and monetize its own compute capacity through the platform its customers already use every day. And the timing lines up with a July 2026 open letter Huang co-signed with Meta, Microsoft, IBM, Mistral, Hugging Face, Mozilla, and Palantir — “Open Weights and American AI Leadership” — arguing that open models strengthen safety, cybersecurity, and national AI sovereignty.
This is Nvidia’s second try. In late 2025, Nvidia offered roughly $500 million for a larger position at a $7 billion valuation; Hugging Face turned it down, wanting to avoid letting any single investor dominate its ownership structure. The dynamic changed this summer, when Delangue approached Huang directly about a full sale rather than another investment round. TechCrunch reported Nvidia was “closing in” on a deal August 26; the next day, The Information broke the roughly $12.9 billion figure and NVDA shares rose about 4%, to approximately $225.96.
Nvidia has reportedly agreed to buy Hugging Face, the popular platform which hosts open-weight AI models, for $12.9 billion, according to reports.
Hugging Face raises $235,000,000 in a Series D round at a $4.5 billion valuation. Nvidia participates alongside other strategic investors — a stake, not a claim on the company.
Nvidia offers roughly $500 million for a larger position at a $7 billion valuation. Hugging Face declines — its leadership doesn't want one investor to dominate the cap table.
Hugging Face co-founder and CEO Clément Delangue raises the idea of a full acquisition with Nvidia CEO Jensen Huang himself, rather than through bankers.
The first public signal that talks had advanced past the exploratory stage.
A roughly $12.9 billion report sends NVDA shares up about 4%, to approximately $225.96.
Nvidia files a Form 8-K with the SEC the same day, putting the total deal value at $12,930,300,000.
Nvidia's blog, TechCrunch, Bloomberg, and a CNBC Squawk Box interview with Delangue all go up. Expected close: first half of 2027.
Hugging Face co-founder and Chief Science Officer Thomas Wolf declined to detail the deal’s terms when Bloomberg Tech asked on September 2, the day the agreement was signed, saying the company has “always had a lot of interest” from multiple parties.
Nvidia is nearing an agreement to acquire Hugging Face... Co-founder Thomas Wolf declined to comment.
By the next morning, the reticence was gone. Delangue told CNBC’s Squawk Box what had changed his calculus over the summer:
“During the summer, I think we realized that Hugging Face and open source AI in general was at the turning point, and that it needed more resources, more scale, more visibility.”
Clément Delangue, co-founder and CEO, Hugging Face · CNBC Squawk Box, Sept. 3, 2026
$NVDA has agreed to buy Hugging Face for $12.9B to own one of AI's biggest open-source distribution layers.
No regulator has said anything public about this deal yet. As of September 3, 2026, neither the FTC, the Department of Justice, nor the European Commission has issued a statement confirming, opening, or even previewing a review of the Nvidia–Hugging Face transaction. What is not in dispute is that a deal this size mandatorily triggers a Hart-Scott-Rodino antitrust filing in the United States, and Nvidia’s dominant position in AI computing hardware makes a formal European Union merger review widely expected once the filing is made. Neither has happened yet.
The question analysts expect any such review to focus on isn’t whether Nvidia can own a hosting platform — it’s whether Nvidia can be trusted to keep that platform neutral. Hugging Face today works the same whether a developer is running models on Nvidia GPUs, AMD hardware, Intel silicon, or the custom chips OpenAI, Google, Amazon, and Anthropic are each building. The concern raised in early coverage of the deal is whether Nvidia ownership could, over time, tilt that hosting layer — through pricing, default integrations, or performance optimization — toward its own hardware and away from the very competitors trying to reduce their dependence on it. That is a question for the pending review, not a finding against Nvidia; nothing here alleges wrongdoing that hasn’t happened.
Nvidia’s own announcement gets ahead of that exact question. Huang wrote that Hugging Face “will remain an open platform for the entire AI ecosystem” and, more specifically, that “NVIDIA compute will not be required to build on or deploy through Hugging Face.” On the origin of the deal, he credited Delangue with initiating it: “I am honored that Clem came to me as he considered the next chapter of Hugging Face and believed NVIDIA would be a great home for the company, its community and the future of open models.”
“Hugging Face will remain an open platform for the entire AI ecosystem. NVIDIA compute will not be required to build on or deploy through Hugging Face.”
Jensen Huang, founder and CEO, Nvidia · official company blog
Not everyone is taking the promise at face value. Forbes contributor Janakiram MSV argued the comparison Nvidia’s own supporters keep reaching for — Microsoft’s 2018 acquisition of GitHub, which stayed genuinely open — doesn’t map cleanly here: “Nvidia is not Microsoft, and Hugging Face is not GitHub,” the piece argues, because Nvidia’s core business is the exact hardware layer Hugging Face is supposed to stay agnostic about, a conflict GitHub’s code-hosting business never had with Microsoft’s cloud unit. The New Stack raised a related concern about platform neutrality more broadly — that even well-intentioned ownership changes can shift a community-run resource’s incentives in ways that are hard to see until they’ve already happened.
By total value, this is Nvidia’s second-largest acquisition on record — a scale marker worth noting alongside the strategy, though secondary to it. The size is one more reason regulators are expected to look closely, even before either has said a word publicly.
Nvidia is turning a three-year-old minority stake into full ownership of the platform that hosts most of the open AI world — and paying $12,930,300,000 to do it. Jensen Huang has publicly promised Hugging Face stays neutral and that Nvidia hardware won’t be a requirement to use it. Nothing forces that promise to hold once the deal closes, and no regulator has yet said whether it agrees the platform will stay that way. The deal isn’t final: it still needs regulatory approval nobody has granted, on a timeline — first half of 2027 — that leaves nearly a year for exactly that question to get answered.
