On October 28, 2026, Reuters reported that the U.S. Department of Justice has spoken with people in the industry to assess how Nvidia's acquisition of Hugging Face could affect competition in the AI market. Hugging Face is a leading platform for publishing, sharing and using AI models, and Nvidia, a supplier of AI chips, plans to bring the model-distribution service under its roof. In its September announcement of the deal, Nvidia said it would preserve an environment in which users can choose other companies' chips and clouds after the acquisition. But a promise to keep options open does not necessarily resolve competition concerns. The report points to a broader question: how would owning a distribution hub for AI models change Nvidia's position in the market?
What Is the DOJ Concerned About in Nvidia's Hugging Face Acquisition?
Reuters' report is based on two anonymous people familiar with the matter. According to them, the DOJ asked about the possibility that Nvidia could gain control of a key model-distribution hub in AI, the risk that competitors could be disadvantaged, and how the acquisition might change Nvidia's bargaining power. The department reportedly also spoke with rivals of Hugging Face. The focus appears to be not simply whether models can still be published and downloaded, but how owning a critical piece of AI development infrastructure could affect competition across the market.
Nvidia's own figures show how much influence Hugging Face has as a foundation for AI development. When the acquisition was announced on September 3, Hugging Face had more than 18 million developers, researchers and creators, and more than 3 million publicly available AI models. More than 200,000 companies used it. It is already a platform where large numbers of developers and models come together, and the DOJ's interest lies in the effect of a leading AI chip company acquiring a service of this scale.
According to Reuters, the DOJ declined to comment, Nvidia also declined to comment, and Hugging Face did not respond to requests for comment. Still, it is a normal part of the process for the DOJ to seek input from industry participants when assessing the effect of a corporate acquisition. Such inquiries do not mean any violation has been found, and the DOJ is not necessarily going to sue to block the deal. At this point, the report says only that authorities are gathering information on the competitive effects.
Reuters also described a separate, ongoing DOJ investigation into Nvidia's business practices. That inquiry reportedly covers the acquisitions of SchedMD and Run:ai, as well as conduct following the technology licensing and talent arrangement Nvidia struck with Groq in 2025. It is also said to be examining whether Nvidia gave priority access to AI chips to customers who bought networking equipment or Groq processors. However, that investigation of business practices should be distinguished from the inquiries about the Hugging Face deal. The two cannot be treated as a single formal merger review.
About $12.9 Billion in Total, With Closing Expected in the First Half of 2027
Nvidia signed a definitive agreement to acquire Hugging Face on September 2, 2026, and announced the deal the next day. The total transaction value given in its official blog post is $12,930.30 million, of which the $12.93 billion cited by Reuters is a rounded figure. The deal has been agreed, but it will not close until required regulatory approvals and other conditions are met. The DOJ's inquiries were therefore reported before closing.
Nvidia explains the breakdown in Item 8.01 of its Form 8-K filed with the U.S. Securities and Exchange Commission (SEC). It says the consideration paid to shareholders is about $11.9 billion, subject to certain adjustments. Separately, Nvidia plans to set aside up to about $1 billion for an equity compensation program aimed at retaining employees after the acquisition. In other words, the transaction includes both the consideration for shareholders and the employee compensation program, and simply adding the approximate amounts will not match the exact total in the official blog post.
Nvidia expects the deal to close in the first half of 2027, provided regulatory approvals and other conditions are satisfied. The post-acquisition operating policies announced in September therefore do not describe measures already in effect under Nvidia's ownership. Depending on the outcome of reviews and approvals, the timing and terms of the deal could change.
What Does Nvidia's Pledge of "Openness" Mean?
In the September 3 official blog post, Nvidia CEO Jensen Huang set out a policy under which Hugging Face users could continue to freely choose AI models and frameworks, as well as the computing infrastructure and cloud they use. Nvidia says that people building models on Hugging Face, or deploying through the service, will not be required to use Nvidia's computing platform. The Form 8-K filed with the SEC likewise states that uploading and downloading of models and datasets will remain possible and that support for other companies' chips will be maintained.
These commitments can be broken into three layers according to what users can choose: models, chips and cloud. The table below maps the statements in the September 3 blog post and the SEC filing to the choices Nvidia said it would maintain after the acquisition. It does not verify the legal force of any contractual terms or how the actual service currently operates.
| What users choose | Policy announced by Nvidia | Source |
|---|---|---|
| AI models | Users can choose which models to use, and uploading and downloading of models and datasets will remain possible | Sept. 3 official blog post; Form 8-K Item 8.01 |
| Chips | Nvidia computing platforms will not be required, and support for other companies' chips will be maintained | Sept. 3 official blog post; Form 8-K Item 8.01 |
| Cloud | Users will be able to choose their cloud and inference service providers | Sept. 3 official blog post |
Separating the layers shows that being able to freely obtain AI models and being able to run them on a chip or cloud of one's choice are different issues. For example, even if a model can be downloaded from Hugging Face, users must separately confirm whether it will run on the other company's chips they want to use. Likewise, being able to choose among multiple cloud providers does not mean each can be used on the same terms.
Nvidia's stated policy is to maintain an environment in which other companies' chips and clouds can be chosen. It is not a guarantee that performance or pricing will be the same whichever provider is used. These are the company's commitments about post-acquisition operations, not the verified results of measures already carried out, and not evidence that regulators have found no competitive problem. It therefore cannot be concluded that the acquisition will have no effect on competition simply because a policy of preserving options has been presented.
The DOJ inquiries reported by Reuters bear on exactly this point. Even if Hugging Face continues to support other companies' products, Nvidia's ownership of a large platform that distributes AI models could still change its relationships with competitors and its bargaining power. Whether users formally retain choices and what effect the platform's owner has on competition need to be evaluated separately.
Will Users Remain Free to Choose Other Companies' Chips and Clouds After the Deal?
For developers and companies that use AI models, access to models is only part of the picture. To run an acquired model in a real service, they must choose compatible software and chips, secure the necessary computing resources and review the terms of cloud or inference services. Even if models can be chosen freely, restrictions on the chips and clouds used to run them would reduce flexibility across development and operations as a whole. In assessing Nvidia's pledge of openness, it is therefore important to confirm whether options are preserved not only for publishing and distributing models but also for the actual operating environment.
When one company provides both a model-distribution platform and computing resources, the conditions under which other companies' products can be used also become a competitive factor, beyond whether support is simply maintained. Pricing, ranking within the service, and supported features and performance, for example, could influence how users choose providers. However, Reuters' report does not confirm that such conditions have changed following Nvidia's acquisition, nor does it report that migration to a specific computing platform has been required. At this stage, potential competitive effects must be distinguished from facts that have actually been confirmed.
The next focus is whether Nvidia can obtain the necessary regulatory approvals ahead of the closing expected in the first half of 2027. Also important is how, after the acquisition, Hugging Face users will be able to keep choosing other companies' chips and clouds and use them in real development and operations. The question is not only whether AI models remain freely available, but also what effect a change in ownership of the platform that distributes them could have on competition across the AI market as a whole.