XNews

NVIDIA Acquires Hugging Face for $12.9 Billion

NVIDIA Acquires Hugging Face for $12.9 Billion

NVIDIA and Hugging Face announced a definitive acquisition agreement for $12,930,300,000, approximately 86x revenue on an ~$150M annualized run rate. The deal was jointly confirmed by CEOs Jensen Huang and Clement Delangue. Hugging Face’s scale includes 18 million developers, 3 million models, 500k datasets, 1 million applications, and 200k enterprise users. Clement Delangue stated that open-source AI needs compute, support, collaboration, and visibility to scale, and described a vision for “100 million AI builders owning, not renting, intelligence.” Jensen Huang emphasized that open weights distribute AI leadership across startups, universities, factories, hospitals, farms, classrooms, and main-street businesses, while also noting NVIDIA’s role as Hugging Face’s top open-source contributor (500+ models, 250+ datasets). The transaction is subject to US HSR antitrust review and EU regulation; completion is expected in 2027. A calculation shared separately indicated that an engineer holding 0.1% equity (granted 2019) could receive ~$9.5M after dilution, a ~120x return. AravSrinivas voiced support, calling the repository vital for public access to AI.

Key insights

Hugging Face Sought Resources to Scale Open-Source AI, Citing Infrastructure Challenges

Hugging Face CEO Clement Delangue explained that after a decade of building the platform, open-source AI is at an inflection point and the community has proven it can supplement or replace closed APIs. Scaling requires compute, support, collaboration, and visibility that the company could not achieve alone, especially with 3 million models hosted and 18 million developers. An earlier security incident was publicly attributed to engineering gaps. The overarching goal is to let 100 million AI builders own intelligence rather than rent it, positioning the deal as a way to secure the necessary resources.

NVIDIA Seeks to Distribute AI Demand via Open Weights to Reduce Dependence on Few Giants

Jensen Huang’s announcement framed the acquisition around keeping AI leadership distributed among enterprises, institutions, and communities—from startups and universities to factories, hospitals, farms, and small businesses. The underlying commercial logic noted: NVIDIA’s largest revenue comes from a handful of closed-model hyper-scalers, and all of them are developing in-house chips. Broad adoption of open-weight models that let organizations deploy AI themselves dilutes that concentration, creating a strategic hedge by diversifying GPU demand across a wider base.

Premium Pricing Reflects Open-Source Platform Value at $12.93B and 86x Revenue

The all-cash acquisition price of $12,930,300,000 represents nearly a 3x increase from Hugging Face’s 2023 Series D valuation of $4.5 billion. On an estimated annualized revenue of ~$150 million, the implied revenue multiple is approximately 86x. These figures underscore the premium attached to a developer-rich platform with 3 million models, 500k datasets, and 200k enterprise customers.

Early Employee Equity Could Yield ~$9.5M, a 120x Retturn

An analysis based on the acquisition price suggests that an engineer who joined in 2019 with 0.1% equity would see their stake worth approximately $9.5 million after seven years of dilution. The equity was originally valued at $80,000, making the return roughly 120x. The calculation assumes the deal closes on reported terms.

Deal Subject to US and EU Antitrust, Expected to Close in 2027

The signed agreement requires approval under the US Hart-Scott-Rodino antitrust process and review by the European Union. The parties and external observers anticipate the review process will extend the closing timeline to 2027.

Open-Source Community Welcomes the Deal as Necessary for Accessible AI

AravSrinivas stated that a repository of open models and tools accessible to the public is essential for AI to remain useful and open to society, and expressed satisfaction that NVIDIA is supporting Hugging Face to fulfill that role.

NVIDIA’s Prior Contributions to Hugging Face Strengthen Its Role as Platform Steard

NVIDIA is already the largest contributor of open-source models and datasets on Hugging Face, having published over 500 models and more than 250 datasets. This track record was cited to address skepticism about a hardware company taking over an open-source community hub and underscores why Clement Delangue previously called NVIDIA the “king of US open-source AI.”

Source posts