Nvidia entered a definitive agreement to acquire Hugging Face on September 2, disclosed in an 8-K filed with the SEC and announced publicly the following day. The consideration is $12,930,300,000, split into roughly $11.9 billion payable to Hugging Face stockholders subject to adjustment and an equity-based retention program of up to about $1.0 billion for employees joining Nvidia. It is expected to close in the first half of 2027 subject to customary conditions including regulatory approvals, per Nvidia's 8-K. Hugging Face carries more than 18 million developers, over 3 million models, 500,000 datasets, and more than 200,000 companies using the platform, per Nvidia.
When we covered the reports on August 28, the Bottom Line said to treat it as reported rather than signed until a filing or a company statement landed. Both have now landed, and the substantive change is that the neutrality question got an explicit answer. Jensen Huang said Hugging Face will remain an open platform for the entire ecosystem, that Nvidia compute will not be required to build or deploy through it, and that the brand stays. Clem Delangue approached Huang rather than the other way around, per CNBC. Those are commitments rather than structural guarantees, and they are the terms on which the deal will be judged.
The regulatory posture is the part worth studying, because Nvidia has spent a year avoiding exactly this. The roughly $20 billion Groq arrangement at the end of 2025 was built as a non-exclusive license plus a talent transfer, a structure that triggered no merger filing in the United States, the European Union, the United Kingdom, or China, and that drew a letter from Senators Warren and Blumenthal in March asking whether the design was intended to skirt premerger review. A straight acquisition of a company has no such path. This one triggers Hart-Scott-Rodino review and a European merger filing, which makes it the first Nvidia transaction of real size to be examined on the merits by agencies that have been watching the company for two years.
Nvidia's argument to regulators, that the platform is a deconcentration force because it spreads open models rather than concentrating them, is a reasonable framing of what Hugging Face has done historically. The question agencies will actually ask is narrower: whether the dominant supplier of AI accelerators gains leverage from owning the distribution point and the telemetry that comes with it, given that the open-weight segment is the one part of the market with no serious custom-silicon escape plan. There is also an operational detail worth holding from our own reporting, which is that Hugging Face infrastructure was the target of the July agent incident, so Nvidia is buying a platform whose security exposure is now a matter of public record. Eighteen months is a long window in which rival silicon vendors have every incentive to fund an alternative hub.
Bottom Line
The filing we told readers to wait for has arrived, and the openness pledges are explicit. The deal now rests on antitrust rather than negotiation, and unlike Groq there is no structure available to route around review, so watch the HSR second-request decision as the first real signal.