Nvidia has agreed to acquire Hugging Face for $12.93 billion, folding the world’s largest open-model repository into the company that already sells most of the compute those models run on. The deal turns a distribution platform into a subsidiary of its biggest supplier. That is the part worth sitting with.
Hugging Face’s numbers explain the price tag as much as the ambition does. The platform serves upward of 18 million developers and carries over three million models, a million applications, and half a million datasets. Nvidia CEO Jensen Huang framed the purchase as a commitment to keep that reach intact rather than restrict it.
“Hugging Face will remain an open platform for the entire AI ecosystem,” Huang wrote in a company blog post announcing the deal. “Nvidia compute will not be required to build on or deploy through Hugging Face.” That is a promise from the new owner, not a term written into any published agreement, and Nvidia has not disclosed how it will be enforced or governed going forward.
Hugging Face CEO Clem Delangue described the sale as a resourcing decision rather than a rescue. In a post on X, he said the platform needed “more compute, more support, more collaboration, and more visibility” to scale further, and that Huang offered to supply exactly that. Delangue had told TechCrunch in July that the company’s growth was pushing it “close to profitability” on its own.
This is not Nvidia’s first attempt to own the platform. It participated in Hugging Face’s $235 million round in 2023 alongside Salesforce Ventures, Google, Amazon, and IBM. According to the Financial Times, Hugging Face turned down a $500 million offer from Nvidia last year. The Information reported last month that the company was generating roughly $150 million in annualized revenue. At $12.93 billion, Nvidia is paying close to 86 times that figure, a multiple that only makes sense if the buyer is pricing the position, not the business.
That position is the real story. Nvidia is the dominant supplier of the chips that train and run AI models, and it now also owns the shelf space where most open models get distributed and discovered. Every rival lab and every competing chipmaker that publishes weights on Hugging Face must now decide what it means to distribute through a competitor’s storefront, and whether an alternative hub is worth building.
Huang has been an outspoken advocate for open-weight models, co-signing a letter arguing they strengthen the United States’ position against rivals such as China. Nvidia has backed that stance with capital: last month’s reported $6 billion arrangement with coding startup Poolside around open models, and, per its own earnings call, upwards of $50 billion committed across frontier labs. Huang has also pointed to open models as infrastructure for autonomous cybersecurity systems, telling analysts that “almost all open models run on Nvidia hardware.”
For any lab or developer that treats Hugging Face as neutral infrastructure, the calculus just changed. Watch whether upload volume from Nvidia’s direct competitors slows in the next two quarters, and whether a credible independent alternative for open-model hosting gets funded in response.
This account is based on TechCrunch’s September 3, 2026 report by Ivan Mehta on Nvidia’s confirmed acquisition of Hugging Face.
