Nvidia agrees to buy Hugging Face for $12.9B, report says

Why it matters

Nvidia is buying the distribution layer for open AI, protecting GPU demand as closed labs develop rival chips. The price also tests whether Hugging Face can remain neutral under a hardware owner.

Nvidia has agreed to acquire Hugging Face for $12.9 billion, The Information reported Wednesday, giving Jensen Huang control of the platform where developers publish, find and deploy much of the open AI market's models and datasets.

https://x.com/ClementDelangue/status/2034984221249298732

The Information attributed the agreement to one person with knowledge of it. The reported transaction would turn Hugging Face co-founders Clement Delangue (@ClementDelangue), Julien Chaumond and Thomas Wolf (@Thom_Wolf) into the sellers of one of AI's most strategically important developer platforms.

Hugging Face began in 2016 with a consumer chatbot before Delangue, Chaumond and Wolf redirected the operation toward machine-learning infrastructure. The founders built Hugging Face into a repository resembling GitHub for AI: developers can host models and datasets, collaborate on projects, optimize models for different hardware and pay for storage, computing and deployment services.

That position explains why Nvidia is willing to pay a software multiple that would look extreme even in the current AI market. Hugging Face was recently producing about $150 million in annualized revenue, according to The Information, up from roughly $100 million a couple of months earlier. Annualized revenue extrapolates a recent month's sales over 12 months; it is not the same as contracted recurring revenue.

The $12.9 billion price is therefore about 86 times that run rate. Delangue told The Information in June that Hugging Face had doubled its number of paying subscribers during the first half of 2026, and he later said Hugging Face was close to profitability. Both measures came from Hugging Face, and neither establishes how much of the recent growth came from repeatable subscriptions versus usage-based compute and storage fees.

The purchase price is also nearly three times the $4.5 billion valuation Hugging Face received in 2023. Salesforce Ventures led that Series D, with Nvidia, Google, Amazon, IBM, Intel, Qualcomm, Sequoia Capital and other investors backing the round. Hugging Face has raised more than $395 million, according to a 2025 announcement from Hugging Face.

The sale puts a price on the neutrality that made Hugging Face useful. Cloud providers, chip designers, AI labs and independent researchers could all distribute models through the same venue without handing control to a direct rival. Nvidia was already an investor and partner, yet ownership gives Nvidia influence over the layer where developers decide which models to download, which inference providers to use and which hardware optimizations become easiest to access.

Delangue had already been moving closer to Nvidia's open-model campaign. In March, he called Nvidia "the new American open-source AI king" and said Hugging Face had passed 15 million AI builders. Hugging Face's sale makes that endorsement considerably more literal.

Nvidia buys the distribution layer

Nvidia's incentive is straightforward. Open models create GPU demand without concentrating the AI market inside a handful of closed-model labs. OpenAI and Anthropic are working to reduce their dependence on Nvidia hardware, while Google sells access to its own tensor processing units. A broader field of model developers gives Nvidia more customers and makes CUDA, Nvidia networking and Nvidia's inference software harder to route around.

Nvidia has reinforced that strategy with its Nemotron models and the Nemotron Coalition, which Nvidia introduced in March with model builders and developer-tool companies including Mistral AI, Reflection AI, LangChain and Thinking Machines Lab. Nvidia distributes Nemotron models through Hugging Face alongside its own services and other model platforms.

Owning Hugging Face gives Nvidia a route to shape open AI before a developer rents a GPU. Model pages, hardware compatibility, deployment tools and hosted inference all sit closer to the developer's initial technical decision than a chip purchase does. The acquisition extends Nvidia's reach from supplying compute into deciding how that compute is discovered and consumed.

Hugging Face also offers Nvidia another path into cloud services. Nvidia previously scaled back DGX Cloud, which involved renting Nvidia-powered capacity from cloud providers and reselling it to enterprises. Hugging Face already charges for compute used to train and run models, allowing Nvidia to revive parts of that strategy through a service developers already use rather than rebuilding distribution under the Nvidia name.

The acquisition follows a wider Nvidia spending campaign around the buyers and users of its hardware. The Information reported that Nvidia holds $99 billion in equity investments and had committed another $25 billion as of the end of July. Nvidia also reported record quarterly revenue of $96.2 billion for the three months ended July 26th, giving Huang ample capacity to buy software, models and distribution while Nvidia's data-center business remains highly profitable.

The neutrality question moves to Nvidia

Hugging Face's technical value comes with a governance problem. Developers publishing models optimized for AMD hardware, Google's processors or other Nvidia alternatives would be storing their work on infrastructure owned by the market's dominant AI accelerator supplier. Nvidia benefits if Hugging Face remains broadly useful, since an overtly Nvidia-only platform would drive developers and model publishers elsewhere. Nvidia also benefits from subtle integration advantages that make its own stack the default.

Regulators have confronted that vertical-control issue with Nvidia before. The Federal Trade Commission sued in 2021 to stop Nvidia's proposed $40 billion purchase of Arm, arguing that Nvidia could restrict rivals' access to critical chip technology. Nvidia abandoned that transaction in 2022. Hugging Face occupies a different market, but its role as shared infrastructure for competing models, clouds and processors gives reviewers another gatekeeper question to examine.

For Delangue, Chaumond and Wolf, the reported deal completes a decade-long pivot from a chatbot into the distribution system beneath open AI. For Nvidia, $12.9 billion buys a strategic counterweight to the closed labs trying to replace its chips, plus a direct line to the developers who determine which models and hardware win next.