Tech giant Nvidia has made a major power move in the artificial intelligence sector, officially announcing the acquisition of Hugging Face for a record-breaking $13 billion. This deal stands as the most expensive acquisition ever made by the semiconductor manufacturer, significantly surpassing its purchase of Mellanox in 2020. This strategic shift marks a dramatic turnaround: just one year ago, the New York-based platform had rejected an investment offer from Nvidia to maintain its independence from the market’s dominant players.
As a cornerstone of "open" AI, Hugging Face currently hosts a massive ecosystem comprising 3 million models, 500,000 datasets, and a community exceeding 18 million developers. Unlike closed, proprietary systems, the models hosted on this platform allow any user to access, modify, and run cutting-edge technologies without being exclusively dependent on cloud giants. With over 200,000 corporate clients, the platform has become an essential distribution channel for integrating AI into sectors as varied as industry, healthcare, and agriculture.
For Nvidia, this colossal investment is driven by a need for diversification and long-term demand security. By acquiring the primary hub for open models, the GPU leader aims to reduce its reliance on a few major clients like OpenAI or Anthropic, while fostering an ecosystem where its chips remain the undisputed hardware standard. Jensen Huang, CEO of Nvidia, is betting on the mass adoption of open models to embed his technological solutions into millions of daily tasks worldwide, thereby strengthening his influence across the entire digital value chain.
The success of this merger will now depend on two critical variables. First, the rigorous scrutiny of antitrust authorities, who will examine the neutrality of a distribution tool that has now become the exclusive property of a hardware manufacturer. Nvidia has already pledged to maintain the open and independent nature of Hugging Face to reassure the market. Second, the continued trust of the developer community will be decisive. As the political debate over the safety of open models rages in the United States, the success of this merger could durably redefine the power balance within global artificial intelligence by the time the deal closes, which is expected in 2027.