NVIDIA has agreed to acquire artificial intelligence platform Hugging Face for $12.93 billion, in a deal that would bring one of the largest open-model communities under NVIDIA’s ownership. NVIDIA CEO Jensen Huang announced the acquisition in a blog post yesterday.

Hugging Face has more than 18 million users, who share over 3 million models, 500,000 datasets and 1 million applications. More than 200,000 companies use the platform. The company said Hugging Face will remain an open platform after the acquisition. Developers will still be able to choose their models, frameworks, cloud providers, and inference services. NVIDIA hardware will not be required to build or deploy through Hugging Face. 

The platform will also continue supporting multiple cloud providers and AI accelerators. NVIDIA said it will use its infrastructure and engineering resources to improve Hugging Face’s reliability, safety, model evaluation, inference, and deployment capabilities.

Why Hugging Face matters to NVIDIA: Stratechery’s Ben Thompson argued that NVIDIA’s interest in Hugging Face makes sense even without a direct financial rationale. He described Hugging Face as a critical hub for open-source and open-weight models, much like GitHub became a central platform for open-source software. Therefore, owning that hub could help NVIDIA shape the future of open AI without forcing developers to use its hardware.

Thompson also argued that NVIDIA benefits when the AI market remains fragmented. Open models give developers more choices over models, infrastructure, and providers. However, NVIDIA’s broadly compatible architecture can still capture demand across that fragmented ecosystem. As a result, supporting open models could protect NVIDIA’s position even as companies develop alternatives to its technology.

The acquisition therefore gives NVIDIA a strategic foothold in open AI. Crucially, Thompson argued that NVIDIA does not need to mandate its chips on Hugging Face. Ensuring that open models continue to thrive could itself strengthen NVIDIA’s long-term position.

Open models already fit NVIDIA’s strategy: Huang had already linked open models to the company’s hardware strategy during its Q2FY27 earnings call. “Nearly all open models run on NVIDIA,” Huang said, citing the company’s global footprint and “fungible” architecture. He added that open models are “skyrocketing in use” alongside closed models. Moreover, Huang said open models are “foundational to just about every AI start-up and every enterprise company around the world”.

The company’s platform sits at the centre of the open-model ecosystem, while NVIDIA provides much of the computing infrastructure beneath it. Huang also said NVIDIA’s “CUDA ecosystem is literally everywhere”, giving it a strong position as developers choose among models and providers. 

Also read: