Nvidia Raises Bet on Open-Source AI With $13B Hugging Face Deal

It’s a match made in momentum trade heaven: AI’s biggest infrastructure provider is about to get its hands on AI’s biggest collection of open-source roadmaps.

On Thursday, Nvidia confirmed reports that it would acquire Hugging Face, a.k.a. “The GitHub of AI,” in a deal valued at a whopping $13 billion. It’s the latest (and perhaps greatest) expression of Nvidia’s strategy to promote an open-source AI ecosystem. “Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty,” Nvidia CEO Jensen Huang said in an X post announcing the deal.

An Offer They Can’t Refuse

Nothing stays the same for long in the AI world. According to sources who spoke to the Financial Times for a story in January, Hugging Face last year rejected a $500 million investment offer from Nvidia, a cash infusion that would’ve come with a $7 billion valuation and amounted to more than the start-up had raised in its more than 10-year history. But the company told the FT it did not want to exist under the influence of a single “dominant” investor. By this summer, the deal tables had turned, and it was Hugging Face that approached Nvidia, Hugging Face CEO Clément Delangue said on CNBC’s Squawk Box on Thursday. 

So what changed? First, the company employed open-source AI models to ward off a swarm of rogue (and proprietary) OpenAI agents that were trying to access the vast Hugging Face repository of open-source AI models, applications and datasets. Then, “we realized that Hugging Face and open source AI in general was at the turning point, and that it needed more, more resources, more scale, more visibility,” Delangue said. Nothing pushes an indie developer into the open arms of Big Tech like a cyberattack ripped straight out of Neuromancer.

For Nvidia, the deal marks another salvo in a simmering cold war with its biggest clients:

• By pushing cheaper open-source models, Nvidia can pull value away from expensive proprietary models like Anthropic’s and OpenAI’s and toward its own hardware, analysts told The Daily Upside. That may dent the resources that major models could use to develop their own in-house chips to replace Nvidia.
• And as its biggest customers increasingly develop their own in-house chip designs, open-source could help Nvidia diversify its customer base. “Open models let startups, businesses, universities and public institutions build on advanced capabilities without training every model from scratch,” Huang said Thursday.

Carolina on My Mind: The Hugging Face acquisition wasn’t Nvidia’s only win this week. Huang also attended a summit at the Chapel Hill campus of the University of North Carolina, where AI industry leaders successfully lobbied representatives from G20 nations to back a light-touch AI regulatory framework dubbed The Carolina Principles. How’s that for a group hug?