Nvidia said on 3 September it had agreed to buy Hugging Face for $12.93 billion, about $11.9 billion to shareholders and up to $1 billion in retention equity for staff who join. The platform hosts more than 3 million models, 500,000 datasets and a million applications for over 18 million developers, and 200,000 companies use it to find and deploy AI. The deal should close in the first half of 2027, subject to regulatory clearance. A sale near this price was first reported over the weekend of 22 August, with no buyer named.
The price is close to three times Hugging Face’s last valuation, set in 2023. Nvidia said the platform would stay open to the whole ecosystem, keep hosting open-weight models from any developer, and support rival clouds and competing chips. It is Nvidia’s second-largest acquisition, behind the roughly $20 billion it paid for Groq’s assets late last year.
Hugging Face is the closest thing open AI has to neutral ground, the default place model weights live regardless of who trained them or what they run on. Ownership by the company that sells most of the training chips changes what that neutrality rests on, even if nothing on the site changes on day one. Rivals that depend on the platform to distribute their own models now do so through a competitor.
Nvidia has not said who polices the openness commitment once the deal closes, or how long it holds. It has not said how the retention pool vests, whether Clement Delangue and the founding team stay, or how regulators in Washington and Brussels will treat the chip leader buying the model hub. The promise of neutrality is on the record. The thing that would hold it there is not.
