Hugging Face turned down a $500 million investment from Nvidia less than a year ago, saying it didn’t want a dominant backer influencing its decisions. Now it’s reportedly close to selling the whole company. According to TechCrunch, Nvidia has agreed to acquire Hugging Face for $12.9 billion, citing The Information’s reporting. Business Insider adds that no agreement has been signed yet and talks could still fall apart, valuing the deal at more than $13 billion. Neither company has responded to requests for comment, and Nvidia’s silence is notable given how quickly the company typically pushes back on inaccurate reports.
Founded in 2016, Hugging Face is the default destination for developers who want to share, discover, or download open-source AI models. It’s not a chipmaker. It’s not a frontier lab. But it sits at the center of the open-source AI ecosystem in a way that few platforms do. And that’s exactly why Nvidia wants it.
The strategic logic here is pretty direct. Nvidia’s chip dominance faces real pressure. OpenAI, Google, Amazon, and Anthropic are all building their own silicon to reduce reliance on Nvidia hardware. A healthy open-source AI ecosystem gives developers and companies alternatives to those closed labs, which keeps a broader portion of the market buying Nvidia GPUs. Nvidia has already spent tens of billions building its own open-source models for exactly this reason. Owning Hugging Face would extend that bet considerably.
There’s also a cloud computing angle. Nvidia reportedly pulled back its DGX Cloud business around a year ago. Hugging Face already lets developers run models using rented compute. Acquiring it hands Nvidia a cloud re-entry point without building from scratch. It also provides a way to offload unused compute capacity, relevant because Nvidia has committed to covering tens of billions in cloud deals for customers who may not use everything they signed up for.
The relationship between the two companies has been building for a while. Hugging Face CEO Clem Delangue has been publicly aligned with Nvidia’s open-source advocacy throughout this year. He appeared on CBS’s “Face the Nation” earlier this month to discuss how Hugging Face used an Nvidia-modified version of a Chinese open-source model to respond to a cyberattack. He also pointed to a letter signed by Nvidia CEO Jensen Huang and 24 other companies, including Hugging Face, urging the U.S. government to support rather than restrict open models. In a July CNBC interview, Delangue warned that China is “clearly dominating” open-source AI, citing the same letter. That’s not the rhetoric of someone keeping Nvidia at arm’s length.
The pricing tells its own story. Hugging Face raised $235 million in 2023 at a $4.5 billion valuation, with Salesforce Ventures leading and Nvidia among the participants. A $12.9 billion acquisition price is nearly a 3x jump on that figure. The Information puts current annual revenue at around $150 million, up from roughly $100 million just two months prior. Delangue told TechCrunch last month the company was getting close to profitability. So $13 billion is a steep multiple, but it’s one that’s hard to walk away from.
The broader context matters here too. AI infrastructure assets are getting swept up fast. Stripe recently acquired OpenRouter, a model-routing startup founded in early 2023, for reportedly more than $7 billion. OpenRouter had been valued at just $1.3 billion during its Series B in May. The gap between private valuations and acquisition prices is compressing quickly, and companies that sit at critical points in the AI stack are attracting outsized offers.
For developers and founders evaluating what this means practically, the key questions are:
- Whether Hugging Face’s open and neutral positioning survives inside a hardware company with obvious commercial incentives
- How Nvidia manages the trust of open-source contributors who chose Hugging Face precisely because it wasn’t controlled by a chip vendor
- Whether competitors like Together AI, Replicate, or even GitHub’s model hub see a window to capture community users wary of the acquisition
This isn’t done yet. But if it closes, it would be one of the most consequential acquisitions in AI infrastructure so far, and a clear signal that the open-source layer of the AI stack is no longer considered a neutral commons.
