August 28, 2026

AIincider

AI News. No Noise. Just Signal.

Nvidia’s $12.9B Hugging Face Deal Faces Real Antitrust Review

3 min read
Nvidia has agreed to buy Hugging Face for $12.9 billion, the first AI deal it cannot structure around a mandatory merger review. Read the breakdown.

Nvidia has reportedly agreed to acquire Hugging Face, the open-source model repository used by more than 13 million developers, for roughly $12.9 billion. It would be the chip giant’s largest purchase ever. It is also the first deal in its recent buying spree that it cannot structure its way around a mandatory antitrust review.

The Deals That Came Before

Over the past nine months Nvidia spent about $27 billion across three transactions: a $20 billion technology license with Groq, a roughly $900 million arrangement with Enfabrica, and a $7 billion license and equity stake in Poolside. Each was framed as a technology license plus talent transfer rather than an outright acquisition, which arguably kept them clear of the Hart-Scott-Rodino premerger notification trigger.

That structure drew attention. Senators Elizabeth Warren and Richard Blumenthal wrote to CEO Jensen Huang in March 2026 asking whether the Groq deal had been built to evade regulators, and FTC Chair Andrew Ferguson said in January that the agency had begun examining whether such arrangements were designed to escape review.

Why the Nvidia Hugging Face Deal Is Different

A straight purchase at $12.9 billion leaves no room for that approach. The parties must file with the FTC and the Department of Justice and observe a waiting period before closing, and a transaction that size clears EU turnover thresholds as well. The Information first reported the agreement, with CNBC, TechCrunch and Fortune matching it within hours. One caveat stands out: no signed contract exists yet, and neither company has commented.

What Regulators Will Examine

The concern is vertical. Nvidia dominates AI accelerators and drew 92 percent of last quarter’s revenue from its data center segment. Hugging Face, by contrast, is deliberately hardware neutral. Its Optimum AMD and Optimum Intel libraries let developers run models on AMD and Intel silicon with no Nvidia GPU involved. Reviewers will ask whether a single owner of both would quietly favor its own chips through search rankings, library maintenance budgets, or early sight of which models are gaining traction.

The price says the rest. Hugging Face generates roughly $150 million a year, so $12.9 billion works out to about 86 times revenue. Nvidia is not buying a software business. It is buying the from_pretrained() call that is wired into hundreds of thousands of production pipelines, at the exact moment Google, Amazon, Microsoft and OpenAI are all shipping chips of their own.

Why It Matters

Hugging Face’s chief executive once described the platform as the Switzerland of AI, precisely because its neutrality is what developers trust. Whether that survives new ownership is now a question for the FTC and Brussels. Combined US and EU review could push a decision well into 2027, and developers with deep Hub dependencies have that long to work out what an alternative would cost them.

Continue Reading…

Leave a Reply