Nvidia nears deal to buy Hugging Face for $12.9 billion

Nvidia has reportedly agreed to buy Hugging Face, the world's most widely used hub for open-source artificial intelligence models, in a deal that would value the company at $12.9 billion, The Information reported on the night of August 26, 2026, according to TechCrunch, which cited a source familiar with the matter. Business Insider, which first reported over the weekend that Hugging Face was fielding takeover interest, said the same night that talks — which would value the company at more than $13 billion — had not yet produced a signed agreement and could still fall apart. TechCrunch reached out to both Nvidia and Hugging Face for comment; neither company had responded by the time of its report, and Nvidia's silence was notable given that the company has moved quickly in the past to correct reports it considered inaccurate.

Founded in New York in 2016 by chief executive Clément Delangue, Julien Chaumond and Thomas Wolf, Hugging Face runs the Hub, a repository where developers publish, download and fine-tune open AI models and datasets. A sale at roughly $12.9 billion would represent close to 90 times the company's estimated annual revenue, according to Portuguese outlet Pplware, and would nearly triple the $4.5 billion valuation Hugging Face reached in a 2023 funding round led by Salesforce Ventures, in which Nvidia itself was among the investors, alongside Alphabet's GV and IBM Ventures.

Why Nvidia wants an open-source foothold

For Nvidia, the acquisition would above all be about protecting its dominance in AI chips, a position that looks increasingly contested even as the company keeps up an aggressive release schedule. OpenAI, Google, Amazon and Anthropic are all now developing their own AI processors to reduce their reliance on Nvidia's hardware. A thriving open-source ecosystem gives their customers an alternative to those closed systems, which in turn keeps a larger share of the market dependent on chips built by Nvidia — a dynamic that has already led the company to pour tens of billions of dollars into building its own open-source AI models.

Hugging Face chief executive Clément Delangue has spent much of 2026 publicly aligned with Nvidia's push for open models, as officials in Washington have debated whether to restrict the distribution of open-weight AI systems. The debate intensified after Chinese developer Moonshot AI released its Kimi K3 model, which matched the benchmark performance of leading American systems while costing far less to train. Earlier this month, on CBS's "Face the Nation," Delangue said Hugging Face had used an Nvidia-modified version of a Chinese open-source model to defend its own infrastructure after a cyberattack, and he pointed to a letter — signed by Nvidia chief executive Jensen Huang and 24 other companies, including Hugging Face — urging the U.S. government to support open models rather than restrict them. In a separate CNBC interview in late July, Delangue made similar points, warning that China is "clearly dominating" open-source AI.

A comeback in cloud computing, and a financial safety net

The deal would also mark something of a return for Nvidia to cloud computing, a business it scaled back roughly a year ago when it pulled back its DGX Cloud offering. Hugging Face already lets developers rent computing power to run their models, and owning the platform, according to The Information, could give Nvidia a way back into that market without starting from scratch. It would also provide a financial cushion: Nvidia has pledged to help cover tens of billions of dollars in cloud-computing commitments made on behalf of its customers, and owning Hugging Face would give it a channel to resell any capacity those customers end up not using.

• $12.9 billion — the price reported by The Information; Business Insider separately puts the valuation at "more than $13 billion"
• $4.5 billion — Hugging Face's valuation in its last funding round, a $235 million raise closed in 2023
• $7 billion — the valuation implied by a $500 million Nvidia investment offer that Hugging Face turned down in late 2025, according to the Financial Times
• $150 million — Hugging Face's estimated annual revenue, up from roughly $100 million two months earlier, according to The Information

Part of a wider consolidation of AI infrastructure

The talks come as Nvidia is also reporting record results of its own, which should not be confused with the price it is said to be offering for Hugging Face. According to Golem.de's account of Nvidia's second-quarter results for its 2027 fiscal year, the company's quarterly revenue came in at $96.2 billion, just under the $100 billion mark, with net profit of nearly $54 billion, driven largely by a data-center division worth about $89 billion that grew 117% year over year. That $96.2 billion figure describes Nvidia's own business for the quarter — it is unrelated to, and roughly seven times larger than, the $12.9 billion to $13 billion price every source consulted attaches to the Hugging Face acquisition itself.

A deal of this size would also fit a broader pattern of consolidation in the layer of companies that sit between AI developers and model providers rather than building frontier models themselves. On August 19, payments company Stripe agreed to acquire OpenRouter, a startup that lets developers switch between AI models depending on cost and task, in a deal reportedly worth more than $7 billion — a sharp jump from the $1.3 billion valuation OpenRouter had reached in a Series B round just three months earlier, in May.

For the millions of developers and companies that rely on the Hugging Face Hub to host, download or fine-tune open models, the outcome of these talks is more than a headline. Hugging Face has positioned itself as a neutral layer, usable regardless of which cloud provider or chipmaker a company favors; an acquisition by Nvidia, already one of its investors, would raise real questions about whether that neutrality holds, how enterprise hosting and compute pricing might change, and whether Nvidia's rivals in AI chips keep the same level of access to the open models their own customers depend on. With no agreement signed and neither company commenting publicly, those questions remain open — but they are the ones worth watching for any business whose products are built on models pulled from the Hub, as the reported talks move toward a conclusion or, as Business Insider cautioned, toward collapse.