Nvidia agreed to acquire Hugging Face for $12.9 billion, giving the world's dominant AI-chip company control of one of the most widely used platforms for sharing and deploying artificial-intelligence models while promising to preserve its open-source character.
Chief Executive Jensen Huang said Wednesday that Hugging Face would continue operating as a platform for developers across the industry rather than becoming a closed Nvidia ecosystem.
"Hugging Face will continue to support open source and open weight models from across the ecosystem, from every model builder," Huang said in a blog post announcing the transaction.
"Together, we will scale Hugging Face's platform, strengthen its infrastructure and expand access to AI for developers and institutions worldwide," he added.
Nvidia shares edged higher following the announcement, rising about 0.25% as of 8:42 a.m. Eastern time. The acquisition, first reported by The Information last week, is Nvidia's second-largest deal on record after its $20 billion purchase of chipmaker Groq.
The transaction brings together a company that supplies much of the computing infrastructure behind modern AI with a platform that has become a central repository for models, datasets and software tools used by developers. Hugging Face allows researchers and companies to share, test and deploy AI systems, making it an important distribution layer in the broader AI ecosystem.
Hugging Face Chief Executive Clément Delangue told CNBC that discussions with Nvidia grew out of concerns that the platform needed substantially more resources as open-source AI matured.
During the summer, "we realized that Hugging Face and open source AI in general was at the turning point, and that it needed more, more resources, more scale, more visibility," Delangue said.
He said he approached Nvidia because he viewed the chipmaker as a "perfect home" for Hugging Face, and negotiations proceeded rapidly.
The purchase price represents a sharp increase from Hugging Face's previous private-market valuation. Nvidia was among the investors in a $235 million financing round in 2023, alongside Salesforce, Google, Amazon and others, that valued the company at $4.5 billion.
At $12.9 billion, Nvidia is paying nearly three times that valuation. TechCrunch has reported that Hugging Face generates roughly $150 million in annual revenue and is approaching profitability.
The deal also marks a reversal from Hugging Face's position only a year earlier. The Financial Times reported that the startup rejected a proposed $500 million investment from Nvidia that would have valued the company at $7 billion because Hugging Face didn't want one shareholder gaining excessive influence.
At the time, Hugging Face had about 13 million users and hosted roughly 2.5 million publicly available AI models, according to the report. Its growth has made the platform increasingly important to developers seeking alternatives to fully proprietary AI systems.
Nvidia and Hugging Face already had significant commercial ties before the acquisition. In 2023, they announced an integration that connected Hugging Face developers with Nvidia's DGX Cloud infrastructure, allowing users to train and customize generative-AI models using Nvidia computing systems.
Buying the platform outright deepens Nvidia's reach beyond the graphics processors that have generated the bulk of its explosive growth during the AI boom. The company has increasingly deployed its cash into investments and acquisitions spanning model developers, cloud infrastructure, networking and other parts of the AI supply chain.
That expansion reflects a broader effort to build a business around the entire AI development stack rather than relying solely on chip sales. Hugging Face gives Nvidia a direct connection to millions of developers at the software and model layer, where decisions about which frameworks, models and computing infrastructure to use are often made.
The acquisition could also raise questions among developers and competing AI companies about whether Hugging Face can remain neutral under Nvidia ownership. The platform hosts models from companies that buy Nvidia hardware but also from developers working with competing chips and software architectures.
Huang's pledge to support models "from every model builder" appears designed to address those concerns at the outset. Maintaining broad participation will be important to preserving Hugging Face's value as a central marketplace and collaboration platform rather than turning it into another distribution channel tied primarily to Nvidia products.
