Nvidia to Acquire Hugging Face for $12.9 Billion, Reshaping the Stack Beneath Small-Business AI
The chipmaker's deal to buy the open-model hub used by 18 million developers concentrates a critical layer of the AI supply chain — with real downstream exposure for the small-team tools founders now rely on for customer acquisition.
Nvidia confirmed on Sept. 3 that it’ll acquire Hugging Face for roughly $12.93 billion, folding the open-model hub used by more than 18 million developers into the same company that already sells most of the silicon those developers train and serve their models on. The structure, per the accompanying 8-K: about $11.9 billion payable to Hugging Face stockholders and up to $1.0 billion in equity retention for employees joining Nvidia. Expected close is the first half of 2027.
It’s Nvidia’s second-largest deal on record, trailing only December’s $20 billion Groq asset purchase. The two acquisitions rhyme. Nvidia is buying the layers immediately above and below its chips: inference silicon on one side, the developer registry and model distribution surface on the other.
Hugging Face hosts more than 3 million models, 500,000 datasets, and 1 million applications. CEO Clément Delangue, who told CNBC’s “Squawk Box” he approached Jensen Huang over the summer because the company “needed more, more resources, more scale, more visibility,” is selling a platform now running at roughly $150 million in annualized revenue. Last year, Hugging Face reportedly turned down a $500 million offer from the same buyer. The revised number is 26 times larger.
Huang’s blog post pledges the platform “will remain an open platform” and that “NVIDIA compute will not be required to build on or deploy through Hugging Face.” The 8-K carries the same commitment in legal register, alongside a regulatory-risk disclosure flagging possible constraints on models “derived from any region, including China.”
For founders at 5-to-30-person companies, the practical read is a supply-chain one. Content generators, outreach tools, and lead-scoring products that pull from Hugging Face-hosted weights now sit downstream of a single hardware-and-platform owner whose economic incentives aren’t yet fully expressed in the platform’s pricing. Auditing which vendors depend on which hosted models, and how their contracts pass through inference costs, is the sort of housekeeping worth doing before close, not after.
The plumbing under small-business AI keeps consolidating faster than the tools on top of it. Stripe’s move to acquire OpenRouter three weeks ago made the same trade at the routing layer. Neutrality is now a promise made by owners, not a property of the market.
Sources
- https://blogs.nvidia.com/blog/nvidia-to-acquire-hugging-face/
- https://www.sec.gov/Archives/edgar/data/1045810/000104581026000078/nvda-20260902.htm
- https://techcrunch.com/2026/09/03/nvidia-confirms-it-will-buy-hugging-face-for-12-9-billion/
- https://www.cnbc.com/2026/09/03/nvidia-agrees-to-buy-hugging-face-for-almost-13-billion-ai-expansion.html
- https://www.bloomberg.com/news/articles/2026-09-03/nvidia-agrees-to-13-billion-deal-for-ai-platform-hugging-face
— END —
