The US Justice Department is examining how Nvidia’s proposed $12.9 billion acquisition of Hugging Face would affect competition in artificial intelligence, turning a deal that Nvidia has framed as a vote of confidence in open AI into a new test of Big Tech’s reach across the AI stack.
Investigators are asking whether the transaction would give Nvidia a critical distribution point for AI models, according to people familiar with the matter. The concern is not simply that Nvidia would own a popular website for sharing models and datasets. It is that control of the place where developers discover, compare and download models could become a source of leverage over the broader AI ecosystem, including rivals that build chips, clouds and software.
Key facts at a glance
- Nvidia agreed in September to buy Hugging Face for $12.9 billion.
- The Justice Department is asking how the deal would affect AI competition.
- Officials want to know whether Nvidia would control a fundamental distribution point for AI models.
- The review is examining whether Nvidia could use Hugging Face to disadvantage rivals.
- Hugging Face is a hub where developers share open AI models and data.
- Nvidia has promised to keep the platform open and expects the deal to close in the first half of 2027.
- The questions are part of a broader, years-long antitrust probe into Nvidia’s conduct.
- Nvidia also faces antitrust scrutiny in France and questions about a separate $20 billion Groq arrangement.
The Justice Department’s questions are running alongside a much broader antitrust examination of Nvidia’s business. That probe began under President Joe Biden and covers the chipmaker’s deals, its stakes in other companies and the way it sells chips, including whether it has put pressure on buyers. Nvidia also faces an antitrust investigation in France, a sign that the scrutiny is not limited to Washington.
The department declined to comment, citing long-standing policy, but said it would continue to prioritize advancement of Super Intelligence, the White House’s term for AI. Nvidia declined to comment. Hugging Face did not respond to a request for comment.
Hugging Face occupies an unusual position in the AI economy. It is best known as a platform where developers publish and download open models, but its role extends beyond a simple repository. The company hosts model cards, datasets, evaluation tools and community discussions that help developers decide which models to use and how to adapt them. For many AI teams, Hugging Face is the first stop when a new open model is released, and often the default place to share a model with the wider world.
That makes the platform a distribution channel rather than just a library. A model that is easy to find, easy to download and well integrated with popular developer tools can spread quickly across startups, research labs and enterprise teams. A platform that controls that discovery process can influence which models gain attention, which technical standards become common and which hardware and cloud services are easiest to use alongside those models.
For Nvidia, owning Hugging Face could strengthen the company’s position at a key point of contact with developers. Nvidia already dominates the market for AI accelerators, and its software ecosystem has become deeply embedded in how AI models are trained and deployed. Hugging Face would add a large community of model builders and users, giving Nvidia more opportunities to align model development with its own hardware and software road map.
The antitrust question is whether that alignment would remain open and competitive or whether it would become a way to steer developers toward Nvidia products and away from rivals. Officials are asking whether the deal would let Nvidia capture a fundamental distribution point for AI, and whether the company could use Hugging Face to hurt competitors. Those questions go to the heart of vertical integration in the AI industry, where control of one layer can affect competition in another.
Why Hugging Face matters to the AI stack
Hugging Face became a central node in the open AI ecosystem by making it easier for developers to share models and collaborate. Its libraries and hosted services are used by researchers, startups and large companies, and its model hub has become a common reference point for new releases. The platform’s value comes from network effects: the more developers use it, the more useful it becomes, and the more useful it becomes, the more developers use it.
That kind of network can be beneficial. It can lower the cost of finding and testing models, encourage interoperability and speed up research. But it can also create chokepoints. If a single company owns the main distribution point for open models, it may be able to set terms for access, prioritize certain models or features, or gather data about how models are used. Those advantages could be especially important in a market where hardware, software and models are increasingly sold together.
Nvidia has said it will keep the platform open after the acquisition. Openness commitments can address some concerns, but regulators often want to know how those promises would be enforced, who would monitor them and what would happen if the market changes. A promise to preserve access is not the same as a structural guarantee of competition, and antitrust officials may examine whether the deal would give Nvidia incentives to favor its own products even while maintaining a public commitment to openness.
The broader antitrust probe
The Hugging Face review is not unfolding in isolation. The Justice Department has been examining Nvidia’s conduct for years, looking at the company’s deals, its investments in other firms and its chip sales practices. One focus has been whether Nvidia has used its market position to pressure buyers or to shape the terms on which AI hardware is sold.
Nvidia’s critics have argued that the company’s dominance in AI chips gives it influence over the entire AI supply chain. Its graphics processing units are widely used to train and run large models, and its software tools are often the default way developers interact with that hardware. That combination can make it difficult for rivals to compete, even if alternative chips offer competitive performance, because developers may be locked into Nvidia’s software ecosystem.
The company’s investments in AI startups and its moves into software and models have also drawn attention. Nvidia is not only a chip supplier. It is increasingly a platform company that provides libraries, frameworks and services used across the AI industry. Each expansion creates new questions about whether Nvidia is competing fairly or using its position in one layer of the stack to gain an advantage in another.
The French antitrust investigation adds another dimension. European regulators have been paying close attention to competition in AI and cloud computing, and France’s probe signals that Nvidia’s business practices are being scrutinized across multiple jurisdictions. For a company of Nvidia’s size, regulatory risk can come from several directions at once, even if the investigations focus on different products or conduct.
The Groq deal and merger-review questions
Last month, the Justice Department was reported to be examining whether Nvidia set up its $20 billion Groq deal to avoid a merger review. Nvidia licensed Groq’s technology and hired its top executives rather than buying the company outright. That structure may have allowed Nvidia to gain access to key technology and talent without triggering the same regulatory process as a traditional acquisition.
The question for regulators is whether such arrangements are functionally mergers in disguise. If a company can obtain the practical benefits of an acquisition through licensing and hiring agreements, the boundaries of antitrust review may need to adapt. The Groq arrangement is separate from the Hugging Face deal, but it illustrates the broader concern that Nvidia is expanding through transactions that may fall outside traditional merger scrutiny.
For Nvidia, the Groq structure may have been a way to move quickly in a fast-changing market. For regulators, it may be a test case for how competition law applies to the AI industry, where talent, data and technology can be combined through many different kinds of agreements. The outcome could influence how other AI companies structure their own deals and how aggressively regulators review them.
Nvidia’s expanding software and model ambitions
Nvidia is the world’s most valuable company, and its grip on AI chips has given it a powerful base from which to move into software and models. The company has invested in AI startups, built tools for developers and expanded its own AI capabilities. Hugging Face would be a major addition to that strategy, giving Nvidia a direct relationship with a large community of open model developers.
Nvidia is also in talks with Reflection AI, a maker of open AI models that Nvidia already backs, about a takeover or a bigger investment. The talks show that Nvidia’s appetite for AI deals extends beyond hardware. As the company moves further into models and software, it will face more questions about whether it is building an integrated platform that competes fairly or an ecosystem that locks in customers and squeezes rivals.
The open-model community has been one of the most dynamic parts of the AI industry. It has produced widely used models, encouraged experimentation and created alternatives to closed systems. If Nvidia owns a central hub for that community, the company could shape the direction of open AI development in ways that go beyond chip sales. That is why the Justice Department’s questions about Hugging Face are not just about one acquisition. They are about who will control the infrastructure of AI discovery and distribution.
What regulators are likely to examine
Antitrust officials will probably look at whether Nvidia could use Hugging Face to favor its own hardware, software or models. They may ask whether the platform could be configured to recommend Nvidia-optimized models, prioritize tools that work best with Nvidia chips or give Nvidia early access to information about new models and developer trends. They may also examine whether Nvidia could degrade access for competitors or make it harder for rival platforms to integrate with Hugging Face.
Another issue is data. Hugging Face collects information about model downloads, usage and community engagement. That data can be valuable for understanding which models are popular and how developers are using AI. If Nvidia can combine that information with its own data about chip usage, it could gain a competitive advantage that is difficult for rivals to match.
Regulators may also consider whether the deal would discourage new entrants. If startups believe that Nvidia controls the main distribution channel for open models, they may be less likely to build competing platforms or competing hardware. The concern is not necessarily that Nvidia would block access outright, but that its ownership of a key platform could shape incentives across the market.
Nvidia’s openness commitments will be part of the discussion. Behavioral remedies, such as promises to maintain access and avoid discrimination, can be difficult to monitor and enforce. Structural remedies, such as requiring divestitures or firewalls, are more intrusive but may address competition concerns more directly. The choice between the two will depend on how regulators assess the risk and how the deal is structured.
The timing of the review is also important. Nvidia expects the Hugging Face deal to close in the first half of 2027, which suggests that the company anticipates a lengthy regulatory process. In the meantime, the Justice Department’s broader probe into Nvidia’s conduct will continue, and other jurisdictions may add their own reviews. The combination of merger scrutiny and conduct investigations creates significant uncertainty for Nvidia’s expansion plans.
The outcome will matter beyond Nvidia. Other AI companies are watching to see how regulators treat vertical deals that combine hardware, software and distribution. If the Justice Department takes a strict view, it could slow a wave of consolidation across the AI stack. If it accepts behavioral commitments, it may signal that companies can expand across layers as long as they preserve access. Either way, the Hugging Face review is likely to become a reference point for how antitrust law applies to the AI economy.
Nvidia has said it expects the Hugging Face deal to close in the first half of 2027, while its talks with Reflection AI continue and its broader antitrust challenges remain unresolved.