NVIDIA agreed to acquire Hugging Face for $12,930,300,000, chief executive Jensen Huang wrote on September 3, 2026. The New York company runs the main public hub for open models, datasets, and apps. Hugging Face posted a matching note that it intends to “join forces with NVIDIA.”

A Form 8-K filed the same week puts the deal on a more precise ledger. NVIDIA signed a definitive agreement on September 2. About $11.9 billion is payable to Hugging Face stockholders, subject to adjustments. Up to about $1.0 billion is reserved for an equity retention program for employees who join NVIDIA. Close is expected in the first half of 2027, after customary conditions and required regulatory approvals.

An open hub, with a written open-platform promise

Huang said more than 18 million developers, researchers, and creators use Hugging Face to share more than 3 million models, 500,000 datasets, and 1 million applications, and that more than 200,000 companies use the platform to find, evaluate, customize, and deploy AI. He named co-founders Clem Delangue, Julien Chaumond, and Thomas Wolf, and said Delangue approached him about the next chapter.

The blog and the 8-K both commit to keeping the hub open. Huang wrote that developers will still pick their models, frameworks, clouds, inference providers, and compute, and that “NVIDIA compute will not be required to build on or deploy through Hugging Face.” The filing adds that the company has committed to let model makers, developers, and users upload and download models and datasets of their choosing, and to keep supporting other silicon vendors.

NVIDIA already treats the hub as a distribution channel. Huang said the company is the largest contributor of open models and data there, with more than 500 models and more than 250 open datasets. The 8-K flags a different risk: governments may restrict open-source model training, release, or transfer, and many widely used open models originate in China before they are downloaded and fine-tuned elsewhere.

Decoded Take

This is NVIDIA buying the town square, not another model lab. The $12.93 billion headline is a cash-and-retention package that keeps Clem’s team on the clock through a 2027 close, which is the tell. Regulators will read the 8-K’s multi-accelerator promise against a chip vendor that already sets the default stack. If the hub stays a place where a Chinese open-weight model, an AMD box, and a European sovereign cloud can still meet, the purchase is infrastructure. If uploads start getting filtered to NVIDIA-friendly recipes, it is a moat with a smile. Watch the closing conditions, whether non-NVIDIA inference endpoints keep first-class placement, and whether the retention pool is enough to stop a talent drain before the deal is even done.