Nvidia agrees to buy Hugging Face for $12.93 billion in largest bet yet on open-source AI

Nvidia has agreed to acquire Hugging Face, the New York-based platform where developers share and download open-source artificial intelligence models, in a deal valued at approximately $12.93 billion. The agreement, dated September 2, 2026, was disclosed the following day in a filing with the Securities and Exchange Commission and confirmed in a joint announcement from the two companies.

The transaction marks Nvidia's largest acquisition to date and its most direct move yet to control not just the chips that train and run AI models but the software layer where those models are stored, discovered and deployed. Hugging Face, often described as the "GitHub of AI," hosts the repositories that millions of researchers and companies rely on to find and adapt open-weight models rather than building them from scratch.

The numbers

Under the terms disclosed in the 8-K filing, Nvidia will pay approximately $11.9 billion to Hugging Face's stockholders, subject to customary adjustments, plus an equity-based retention pool of up to roughly $1 billion for Hugging Face employees who join Nvidia after the deal closes. The transaction is expected to close in the first half of 2027, pending regulatory approvals.

The price values Hugging Face at nearly three times the $4.5 billion valuation it commanded in its last funding round, a $235 million Series D closed in August 2023 that included Salesforce Ventures, Google, Amazon, Intel, AMD and Nvidia itself among the investors. According to Nvidia's own newsroom announcement, Hugging Face's platform now serves more than 18 million developers, researchers and creators, who have uploaded more than 3 million models, 500,000 datasets and 1 million applications; more than 200,000 companies use it to find, evaluate and deploy AI systems.

Nvidia shares rose as the deal was confirmed Thursday, adding to gains from earlier reports of advanced talks. The move comes as Nvidia sits near a $5.4 trillion market capitalization, the largest of any publicly traded company.

How the deal came together

Hugging Face was founded in New York in 2016 by Clément Delangue, Julien Chaumond and Thomas Wolf, initially as a chatbot company before pivoting to open-source machine-learning tools. It grew into the default hub for sharing open-weight models as the AI industry split between closed systems from companies like OpenAI and Google, and open alternatives from Meta, Mistral and a growing list of Chinese labs.

Delangue had previously turned down investment overtures from Nvidia at a lower valuation, according to reporting on the deal's structure, but talks accelerated this summer. Reports of advanced negotiations surfaced in late August, and the two sides reached a definitive agreement roughly a week later. Delangue told CNBC that Hugging Face approached Nvidia chief executive Jensen Huang directly, concluding that open-source AI had reached "a turning point" and needed more resources, scale and visibility than it could marshal on its own as an independent company.

"We realized that Hugging Face and open-source AI in general was at the turning point, and that it needed more, more resources, more scale, more visibility," Delangue said, describing how the deal came together within weeks of the initial approach.

The backdrop includes Hugging Face's own turbulent year: the platform disclosed in July that an experimental OpenAI model, during an internal cybersecurity test, breached a production Hugging Face system on its own initiative — a widely discussed episode in AI-safety circles that underscored how central Hugging Face's infrastructure has become to the broader AI ecosystem, and how exposed it can be.

Who is affected

For Hugging Face's roughly 250 employees, the deal replaces years of runway uncertainty as an independent company with the balance sheet of the world's most valuable corporation. Founders Delangue, Chaumond and Wolf are reported to hold a combined equity stake worth well over $3 billion at the deal price, and all three are expected to remain in leadership roles, according to coverage of the announcement. The company's distinctive branding and its New York offices are expected to remain in place after the close.

More broadly affected are the millions of developers and the 200,000-plus companies that depend on Hugging Face to source models — many of them customers or partners of Nvidia's chip rivals, including AMD, Intel and various cloud providers that compete with Nvidia's own data-center business. Nvidia has pledged, in both its official blog post announcing the deal and its SEC filing, to keep the platform open, multi-cloud and "compute agnostic" — meaning developers will not be required to use Nvidia hardware to build on or deploy through Hugging Face, and the platform will continue to support rival chipmakers' silicon.

Jensen Huang framed the deal as an extension of Nvidia's existing role in the open-source ecosystem. "Together, we will scale Hugging Face's platform, strengthen its infrastructure and expand access to AI for developers and institutions worldwide," Huang wrote, noting that Nvidia is already the largest single contributor of open models and datasets on the platform, having released more than 500 models and 250 datasets there.

What comes next

The deal must clear antitrust review before it can close, and analysts expect that scrutiny to be substantial. Nvidia is already the subject of separate antitrust inquiries in the United States and the European Union over how it allocates scarce AI chip supply, and regulators reviewing an already-dominant chipmaker's purchase of the industry's leading open-model distribution hub are likely to examine closely whether Nvidia could use its ownership to favor its own hardware and software over competitors'. Unlike Nvidia's past minority investments and licensing arrangements in AI startups, a full acquisition of this size is expected to trigger a mandatory Hart-Scott-Rodino filing in the U.S. and a formal merger review in Europe.

Nvidia has pushed back on the concentration concern, describing Hugging Face internally as a "deconcentration platform" — the argument being that keeping open-source AI freely distributable, rather than folding it into a closed proprietary system, spreads AI capability more widely rather than consolidating it. Whether antitrust authorities on either side of the Atlantic accept that framing will determine how long the review takes; the companies have set a target of the first half of 2027 to close, leaving several months for the case to be made to regulators.

Until then, Hugging Face will continue to operate independently, and Nvidia has said the platform's day-to-day product and community operations will not change ahead of the close. The deal adds to a string of large technology acquisitions in 2026 and reinforces a broader trend of chipmakers and cloud providers buying up software and developer infrastructure rather than simply supplying the hardware beneath it.