Nvidia Is Buying Hugging Face for $12.93 Billion—And Promises Its Open AI Hub Will Stay Open

A company built around sharing artificial-intelligence models is changing hands in a deal that puts one of the field’s most important public meeting places inside the world’s dominant AI-chip maker.

Nvidia said Thursday, September 3, that it will acquire Hugging Face for exactly $12,930,300,000. The nearly $13 billion agreement brings the widely used model-and-dataset hub under Nvidia at a moment when chipmakers, cloud providers and AI labs are racing to control more of the technology stack.

The central question is bigger than the price: Can a platform used by developers across competing hardware and cloud ecosystems remain meaningfully open when it belongs to one of those ecosystems’ most powerful suppliers?

What Nvidia is buying

Hugging Face is far more than a download site. Developers use its Hub to publish, test and collaborate on machine-learning models and datasets, while its software libraries and cloud services have become common building blocks for AI projects.

The company was founded in 2016 by French entrepreneurs Clément Delangue, Julien Chaumond and Thomas Wolf. It began with a consumer chatbot before pivoting toward open machine learning. Today, the platform contains more than 3 million models and 500,000 datasets and serves roughly 18 million developers, according to the Financial Times.

That scale makes Hugging Face a distribution layer, a collaboration network and a map of what developers are actually building. For Nvidia, those relationships and signals may be nearly as valuable as any individual piece of software.

The promise that matters most

Nvidia CEO Jensen Huang said the companies intend to strengthen Hugging Face’s infrastructure, expand access and keep the service open. Crucially, Nvidia says developers will remain free to use the platform without choosing Nvidia hardware.

That promise addresses the most obvious concern surrounding the deal. Hugging Face’s community includes researchers, startups and major companies whose products can compete with Nvidia or run on rival processors. The platform’s usefulness depends partly on the belief that it is a neutral place to discover and share AI work.

According to Reuters, Hugging Face’s backers include Intel, AMD and Amazon—three companies with their own interests in chips or cloud computing. Nvidia’s pledge will therefore be judged in practical details: which tools get the best integration, how competing compute options are presented, and whether developers continue to see equal access over time.

Why Nvidia wants a model hub

Nvidia built its AI leadership by selling the graphics processors and systems used to train and run advanced models. But several of its largest customers are designing chips of their own, seeking lower costs and less dependence on a single supplier.

Owning Hugging Face gives Nvidia a powerful position farther up the stack, where models are discovered, adapted and deployed. It could make Nvidia’s software and infrastructure easier to use without forcing the company to win every contest solely at the chip level.

The strategy also aligns Nvidia with rising demand for “open-weight” models—systems whose trained parameters can be downloaded and modified. Lower-cost models from companies including China’s DeepSeek and Z.ai have intensified competition and shown how quickly capable alternatives can spread when their weights are broadly available.

This does not mean every model on Hugging Face is fully open source in the traditional software sense. Licenses and access conditions vary. But the platform has become the best-known marketplace and workshop for AI that can be inspected, adapted or run outside a single vendor’s closed service.

A premium price for influence

The purchase price is almost three times Hugging Face’s reported $4.5 billion valuation in 2023. The Verge reports that the startup generates about $150 million in annual revenue, underscoring that Nvidia is paying for strategic reach and future potential—not merely today’s sales.

Reuters reports that the transaction includes $11.9 billion for investors and about $1 billion in equity incentives for employees. The structure highlights another challenge: retaining the engineers and community leaders whose trust made the platform valuable in the first place.

What developers should watch next

The deal is expected to face regulatory review before closing. Regulators may examine whether combining the leading AI-chip supplier with a central model-distribution platform could disadvantage hardware rivals or steer customers toward Nvidia products.

For developers, nothing changes overnight. The important tests will arrive after the corporate announcement: whether Hugging Face’s branding and governance remain distinct, whether rival hardware receives comparable support, and whether policies around models, datasets and access change.

Nvidia’s official announcement listing makes the commitment to openness part of the deal’s public rationale. Preserving that openness is also a business necessity. The platform is worth nearly $13 billion precisely because millions of developers chose it as shared ground.

If Nvidia keeps that ground genuinely open, the acquisition could give Hugging Face more infrastructure and a longer runway. If the hub begins to feel like a sales channel for one hardware company, the community that created its value will have strong reasons to build somewhere else.