How much is Nvidia paying for Hugging Face?

Nvidia has agreed to acquire Hugging Face for $12,930,300,000, announced September 3, 2026. It's Nvidia's second-largest acquisition on record, after its $20 billion purchase of assets from Groq.

Has the acquisition officially closed?

As of this writing, no. The deal has been announced and agreed to, but it hasn't closed, and Hugging Face continues to operate independently in the meantime.

Is this deal facing antitrust review?

Analysts and rival chipmakers expect regulators in the US, EU, and potentially the UK to review the deal for vertical foreclosure or self-preferencing risk, similar in spirit to the scrutiny that sank Nvidia's earlier attempt to buy Arm.

Did Hugging Face turn down an earlier offer from Nvidia?

Hugging Face reportedly turned down a $500 million offer from Nvidia in late 2025, before agreeing to the current $12.93 billion deal less than a year later.

Nvidia to Buy Hugging Face for $12.9B: What It Means

Q: Will I need Nvidia hardware to keep using Hugging Face?

Nvidia says no. Its stated terms are that developers can keep choosing their own models, frameworks, clouds, and computing platforms, and that Nvidia compute won't be required to build on or deploy through Hugging Face.

Nvidia has agreed to buy Hugging Face, the platform that hosts the open-source models most developers pull from, for $12.93 billion. It's Nvidia's second-biggest acquisition ever, and it puts the chip company in charge of the hub 18 million developers use to find, share, and deploy AI models — which is exactly why the deal is drawing antitrust attention before it's even closed.

Key Takeaways

Nvidia announced on September 3, 2026 that it's acquiring Hugging Face for $12,930,300,000, roughly 26x the $500 million offer Hugging Face reportedly turned down in late 2025.
Hugging Face hosts more than 3 million models, 500,000 datasets, and 1 million applications, used by over 18 million developers and 200,000-plus companies, according to Nvidia's own figures.
Nvidia says Hugging Face stays hardware- and cloud-agnostic — no Nvidia compute requirement to build on or deploy through the platform.
Analysts and rival chipmakers expect antitrust scrutiny in the US, EU, and possibly the UK, centered on whether Nvidia's ownership can stay neutral over time.
The deal hasn't closed yet, and Hugging Face continues to operate as it did before the announcement.

What exactly did Nvidia announce?

In a September 3 blog post, Nvidia CEO Jensen Huang said the company has agreed to acquire Hugging Face for $12,930,300,000. The announcement followed weeks of reporting that the two companies were in talks, after Hugging Face had reportedly rejected a $500 million offer from Nvidia in late 2025. Huang's post frames the deal around scale: Nvidia's infrastructure and engineering, paired with Hugging Face's existing community of more than 18 million developers, researchers, and creators sharing over 3 million models, 500,000 datasets, and 1 million applications. Hugging Face CEO Clément Delangue said on X that he approached Huang directly, arguing the platform needed more compute, support, and visibility to keep growing at scale.

Why would Nvidia want Hugging Face?

Huang has spent the past year publicly championing open-weight models, including co-authoring a letter arguing that open weights broaden AI access and keep leadership distributed rather than concentrated in a handful of labs. Nvidia is already the largest single contributor of open models and data to Hugging Face, with more than 500 models and 250 datasets released there. Owning the platform outright gives Nvidia a direct hand in how the next generation of open models gets built, evaluated, and shipped — and, as TechCrunch has noted, a channel to sell spare compute capacity to enterprise customers alongside Hugging Face's existing offering. Hugging Face itself was reportedly closing in on profitability off roughly $150 million in annualized revenue before the deal, so this isn't a rescue — it's Nvidia buying a growing, influential platform at the center of the open-model ecosystem.

Will Hugging Face stay neutral for non-Nvidia hardware?

That's the question actually worth watching. Nvidia's official line is unambiguous: developers keep choosing their own models, frameworks, clouds, and computing platforms, and Nvidia compute won't be required to use Hugging Face. Nvidia enterprise computing GM Justin Boitano has described the deal as building a "deconcentration platform" that counterbalances the market power concentrated in a few proprietary model APIs — which doubles as Nvidia's opening argument against antitrust concerns.

The skepticism isn't really about day-one behavior, though. Analysts have pointed out that Hugging Face's value comes specifically from its neutrality, so Nvidia has every reason to preserve openness at first. The longer-term risk people flag is subtler: quantization formats, serving optimizations, or new library features that quietly ship Nvidia-first, with AMD, Intel, and other backends catching up months later. None of that requires bad intent — ordinary roadmap prioritization inside an Nvidia-owned Hugging Face could produce the same effect regulators normally associate with self-preferencing.

What's the antitrust risk, and when does the deal close?

This isn't Nvidia's first brush with regulators over a platform acquisition. Its 2020 attempt to buy Arm for $40 billion collapsed after the FTC, UK, EU, and China all pushed back, arguing that Arm's neutral licensing position — dubbed the "Switzerland of hardware" — shouldn't fall under one chipmaker's control. More recently, Nvidia's Run:ai acquisition, roughly a tenth the size of the Hugging Face deal, drew an EU merger referral that Nvidia fought before the deal ultimately went through. Hugging Face is an order of magnitude bigger and far more visible to developers worldwide, so expect information requests and preliminary review from US and EU regulators at minimum, with rival chipmakers like AMD and Intel among those likely to raise concerns during any review process.

As of this writing, the deal is signed but not closed — Hugging Face continues to operate independently in the meantime. Some analyst notes have floated a close in the first half of 2027 given the expected regulatory timeline, though Nvidia hasn't published an official closing date.

Is this Nvidia's GitHub moment?

The comparison people keep reaching for is Microsoft's $7.5 billion purchase of GitHub in 2018. Microsoft wasn't buying the code hosted there — it was buying the place where a huge slice of the world's developers already gathered, and GitHub has stayed genuinely useful to non-Microsoft stacks since, even as it funneled plenty of engineers toward Azure over time. Hugging Face occupies a similar spot for AI models and datasets. Whether Nvidia's ownership plays out the same way, or tilts the platform toward CUDA-optimized paths over the next few years, is the thing to actually watch — not the press release language, which says exactly what you'd expect a company in Nvidia's position to say.

Wrapping Up

Nothing changes for how you use Hugging Face today — the deal hasn't closed, and Nvidia's stated terms keep it hardware-agnostic. The practical move for teams with real dependencies on the platform is to keep an eye on regulatory filings over the next several months and pay attention to which backends new model releases and optimizations ship for first, since that's a far more reliable signal of where things are headed than any blog post.