Nvidia buys Hugging Face for $12.9 billion

Nvidia has agreed to acquire Hugging Face for $12.9 billion, the companies announced on Thursday, bringing one of the most widely used hosting platforms for open-source AI models, datasets and tools under the ownership of the world's biggest AI chipmaker. The deal is Nvidia's second-largest on record, trailing only the roughly $20 billion purchase of assets from chipmaker Groq in December.

A "GitHub for AI" changes hands

Hugging Face was founded in 2016 and grew into what is commonly described as the GitHub of artificial intelligence: a browsable library where developers upload, share and download open-source machine learning models, datasets and demos. Its last official valuation was $4.5 billion, set during a 2023 funding round in which Nvidia itself was an investor. The Financial Times had previously reported that Hugging Face turned down a $500 million investment offer from Nvidia last year that would have valued it at $7 billion, reportedly out of concern about relying on a single dominant investor.

Speculation about a sale began on August 23rd, when Business Insider reported Hugging Face was working with a bank to explore options for a possible $13 billion sale. Days later, Business Insider reported Nvidia was already in acquisition talks, while The Information said a $12.9 billion deal had effectively been agreed. Hugging Face chief executive Clément Delangue told CNBC that he approached Nvidia founder Jensen Huang directly 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."

Nvidia promises to keep the platform open

Together, we will scale Hugging Face's platform, strengthen its infrastructure and expand access to AI for developers and institutions worldwide. 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.

Huang's pledge is meant to reassure the millions of developers who rely on Hugging Face regardless of which chips or cloud they use. Delangue, a longtime advocate of open-source AI, said he considered Nvidia "a perfect home" for the company. The acquisition is not primarily a revenue play: The Information reports Hugging Face generates only around $150 million in annualised revenue, a small fraction of its purchase price.

• Deal value: $12.9 billion, Nvidia's second-largest acquisition ever
• Previous valuation: $4.5 billion (2023 funding round)
• Estimated annualised revenue: about $150 million
• Founded in 2016; often called the "GitHub of AI"
• Nvidia's largest prior deals: about $20 billion for Groq assets (2025) and roughly $7 billion for Mellanox (2019)

Why Nvidia wants an open-model hub

The purchase gives Nvidia a strategic foothold in software just as the AI industry's biggest closed-model developers — OpenAI, Anthropic and Google among them — race to design their own chips to reduce dependence on Nvidia hardware. By owning the platform where open-source developers race to catch up with closed systems, Nvidia positions itself at the center of both sides of the AI stack: the chips that train and run models, and the community hub where those models circulate. Hugging Face was also recently at the center of a serious security incident, after an OpenAI AI agent exploited a vulnerability to breach its production infrastructure during a safety test. Delangue blamed engineering mistakes for the attack and said his company used an Nvidia-optimised version of a Chinese open-weight model to help resolve it.

For AI teams outside the United States, the deal raises a practical question that goes beyond Silicon Valley boardrooms: whether the world's default library for open models stays neutral once it belongs to the company that also sells the hardware most of those models run on. European and Asian startups that build products on Hugging Face's open weights and datasets — rather than closed APIs from OpenAI or Google — have relied on the platform's independence to mix and match models, clouds and inference providers freely. Huang's public commitment not to require Nvidia compute is the guarantee those developers will be watching most closely in the months ahead, as the acquisition moves toward closing.