NVIDIA to acquire Hugging Face for $12.93 billion

NVIDIA has agreed to acquire Hugging Face, the open-model hosting platform that has become the default distribution point for open-weight AI, in a deal worth $12.93 billion. The announcement came September 3, 2026, via a post on NVIDIA's own blog, quoting CEO Jensen Huang directly on the terms and intent of the deal.

What's new

NVIDIA confirmed the exact deal value down to the dollar: "NVIDIA has agreed to acquire Hugging Face for $12,930,300,000. Together, we will scale Hugging Face's platform, strengthen its infrastructure and expand access to AI for developers and institutions worldwide," Huang said in the announcement. The companies did not disclose a closing timeline or list specific closing conditions in the post.

NVIDIA committed to preserving Hugging Face's neutrality as a hosting platform rather than folding it into NVIDIA's own stack. The announcement states plainly: "Hugging Face will remain an open platform for the entire AI ecosystem. Developers will choose the models they want, the frameworks they want, the clouds and inference service providers they want and the computing platforms they want." NVIDIA also pledged continued support for third-party open releases: "Hugging Face will continue to support open source and open weight models from across the ecosystem, from every model builder." Branding stays intact too — the post specifically notes Hugging Face keeps "their same iconic 🤗 brand."

Context

Hugging Face has spent the past several years becoming the closest thing the AI industry has to a neutral clearinghouse: model weights, datasets, Spaces demos, and the transformers and diffusers libraries that much of the open-source AI tooling ecosystem is built on. Nearly every major open-weight lab — Meta, Mistral, DeepSeek, Qwen, and dozens of smaller players — publishes there first. That neutrality is precisely what has made Hugging Face valuable as infrastructure, and it's also the reason an acquisition by any single AI vendor was always going to draw scrutiny.

NVIDIA, for its part, has been steadily extending its reach beyond chips and into the AI software layer — inference runtimes, enterprise AI stacks (Nemotron), and now, with this deal, distribution. The company has also been an active Hugging Face partner and investor in the AI ecosystem more broadly, including a $3.5 billion investment in MediaTek disclosed just days earlier.

Why it matters

This is the largest acquisition in the AI infrastructure layer this year, and it hands NVIDIA a chokepoint that sits above the hardware layer it already dominates: the place where open models actually get published, downloaded, and served. NVIDIA's explicit public commitment to keep the platform framework-neutral and cloud-agnostic is a direct answer to the obvious worry — that the company would use Hugging Face to steer developers toward NVIDIA-only tooling or CUDA-locked inference paths. Whether that commitment holds under commercial pressure, once NVIDIA owns the rails the rest of the open-source ecosystem depends on, is the question every lab publishing on Hugging Face will now be watching. For a company already vertically integrated from silicon (Blackwell/Rubin) through inference software (TensorRT, Nemotron) up to enterprise deployment (with partners like CrowdStrike), owning the model hub closes one of the last remaining gaps between NVIDIA and the entire AI stack.