Nvidia has agreed to acquire Hugging Face, the leading hub for open AI models, datasets, and applications, for about $12.9 billion. The chipmaker framed the purchase as a way to deepen its role in the open-model ecosystem while pledging that the platform will remain open and multicloud after closing.
Hugging Face hosts millions of developers, millions of models, hundreds of thousands of datasets, and more than a million applications, with hundreds of thousands of companies using the service. Putting that marketplace inside Nvidia is a landmark consolidation: the company that sells most of the world’s AI accelerators would also own the default distribution layer for open weights.
The open-model community reacted with anxiety as well as relief. Nvidia promised users would still be able to choose competing models, clouds, and hardware. That pledge is now the test of the deal. If Hugging Face stays a neutral bazaar, the acquisition looks like infrastructure. If it quietly favors Nvidia stacks, it looks like vertical lock-in.
The New York Times and other outlets described the purchase as another sign that Nvidia is acting as Silicon Valley’s “central banker” for AI—financing, acquiring, and underwriting the platforms that sit on top of its chips.
Key takeaway — The $12.9 billion Hugging Face deal folds the industry’s main open-model marketplace into Nvidia. Whether the platform stays hardware-neutral will decide if this is stewardship or capture.
Image: Hugging Face (SVG hero could not be imported into the WordPress media library). Sources: Nvidia blog; Bloomberg; New York Times; Reuters; AI Briefing.
