Nvidia’s announcement on September 3 that it will pay $12.93 billion for Hugging Face marks a deliberate shift in strategy for a company that has spent three years selling the picks and shovels of the artificial intelligence boom rather than the platforms built on top of them. The deal, which includes an equity-based retention pool worth as much as $1 billion for Hugging Face staff who join Nvidia, is the chipmaker’s second-largest acquisition on record. It values a company that started as a chatbot app for teenagers at a premium few would have predicted when it pivoted to open-source machine learning tools less than a decade ago.
The scale of what Nvidia is buying explains the price. More than 18 million developers, researchers and hobbyists use Hugging Face to share and download models, and the platform now hosts more than three million of them, according to figures released alongside the deal. Over 200,000 companies rely on it to discover and deploy AI systems, effectively making it the default distribution layer for open-weight models in the same way GitHub became the default home for source code. Owning that layer gives Nvidia visibility into which architectures are gaining traction before they show up in chip orders.
Chief executive Jensen Huang framed the acquisition around sovereignty and safety rather than distribution economics, arguing that open models “strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty” for developers, universities and governments building their own AI capacity. That framing matters commercially: national governments wary of depending on closed, foreign-controlled models have become an important growth market for Nvidia’s hardware, and control of the leading open-model hub reinforces the case that its stack is neutral infrastructure rather than a proprietary walled garden.
The transaction, expected to close in the first half of 2027 pending regulatory approval, will invite scrutiny from antitrust authorities in Washington and Brussels already wary of Nvidia’s dominance in AI accelerators. Critics will ask whether a company that controls the dominant chips, a leading cloud software stack through its enterprise offerings, and now the largest open-model marketplace has become too central a chokepoint for a technology increasingly treated as critical infrastructure. Nvidia’s rivals, including cloud providers building their own silicon, will be watching closely to see whether access to Hugging Face’s ecosystem remains as open as Huang has promised. Nvidia shares moved only modestly on the announcement, a sign investors are still weighing whether the chipmaker should be diversifying into software platforms at all.