- Nvidia confirmed on September 3, 2026 it will acquire Hugging Face for $12.93 billion, ending weeks of rumors.
- Hugging Face hosts more than 3 million models and is used by over 18 million developers, per Nvidia.
- The deal puts the main hub of open-source AI under the industry’s dominant chipmaker.
- It raises questions about the neutrality of a platform that hosts rivals’ models.
What Happened
Nvidia confirmed on September 3, 2026 that it has agreed to acquire Hugging Face for $12.93 billion, as first reported by TechCrunch. “After weeks of swirling rumors, Nvidia confirmed today that it has acquired Hugging Face,” the outlet reported. Nvidia said Hugging Face hosts more than 3 million models and is used by over 18 million developers, making it the default distribution point for open-weight AI.
Why It Matters
Hugging Face is, in effect, the GitHub of machine learning — the neutral ground where Meta’s Llama, Alibaba’s Qwen, Mistral, DeepSeek, and thousands of smaller labs publish weights and datasets. Wired framed the deal as “a $12.9 billion bet on open-source AI.” Bringing that hub inside Nvidia, which already supplies the GPUs most of those models train and run on, extends the company’s control from silicon up through the software and community layer.
It also caps a run of consolidation across the AI stack. In recent months SpaceX acquired the coding startup Cursor and Stripe bought the model-routing firm OpenRouter — deals that, like this one, fold widely used developer infrastructure into larger strategic owners. For Nvidia, a roughly $13 billion outlay is modest against a market capitalization measured in the trillions, and buys the distribution layer for the open-model movement that increasingly drives GPU demand.
Technical Details
Beyond the model registry, Hugging Face maintains widely used open-source libraries — Transformers, Diffusers, and the Hub APIs — that are embedded in a large share of AI development workflows. Nvidia’s stated figures (3 million models, 18 million developers) quantify the reach it is buying. The strategic logic is distribution: optimizing those libraries and models for Nvidia hardware by default would deepen the moat around CUDA, the software layer that already locks developers to Nvidia GPUs.
Who’s Affected
Open-model labs that rely on Hugging Face for neutral distribution — including direct Nvidia customers and competitors — now depend on a platform owned by a chipmaker with its own commercial interests. Hugging Face co-founder Thomas Wolf, discussing the deal on Bloomberg Tech, addressed the rationale for combining. Developers gain deeper Nvidia integration; rival hardware makers such as AMD and cloud silicon teams have the most to lose if the hub tilts toward Nvidia optimization.
What’s Next
The deal’s completion will depend on regulatory review, given Nvidia’s dominance in AI accelerators and the platform’s central role in the ecosystem. The signal to watch is governance: whether Nvidia commits to keeping Hugging Face hardware-neutral, or whether competing accelerators gradually become second-class on the platform.
