Nvidia Advances in Its Acquisition of Hugging Face

Nvidia has come to a deal to acquire Hugging Face for $12.9 billion, according to a report from The Information on Wednesday night, citing insider sources. Business Insider revealed over the weekend that Hugging Face was considering acquisition offers, mentioning that although the negotiations — which would value the company at over $13 billion — have not yet resulted in a finalized agreement, there remains a possibility they could fall through.

TechCrunch previously reached out to both Nvidia and Hugging Face for comments, but neither has yet responded. Nvidia’s lack of comment is noteworthy, considering the company’s track record of swiftly addressing erroneous media reports.

It seems almost unavoidable. Founded in 2016, Hugging Face has established itself as a premier platform for developers to share and download open-source AI models. Acquiring Hugging Face would grant Nvidia a substantial foothold in the open-source AI space, particularly as developers strive to connect with closed AI systems from firms like Anthropic and OpenAI.

What drives Nvidia’s interest in this acquisition? The main motive is to protect its leading position in the AI chip market, which appears increasingly at risk given that major closed-source AI laboratories (OpenAI, Google, Amazon, and Anthropic) are actively crafting their own AI chips to lessen reliance on Nvidia. A strong ecosystem of open-source AI models provides customers alternatives to these closed labs, ultimately ensuring a larger segment of the market continues to rely on Nvidia’s hardware. This is also why Nvidia has invested billions in developing its own open-source AI models.

Is it surprising that Hugging Face’s future as an independent organization appears uncertain? Not at all. Hugging Face CEO Clem Delangue has openly aligned himself with Nvidia’s open-source efforts this year. This comes amid ongoing talks, as Washington officials consider implementing restrictions on open-weight models. With Chinese companies like Moonshot AI launching competitive systems against top U.S. models that are also cheaper to operate, concerns regarding competitiveness and national security have escalated, with critics such as White House advisor David Sacks referring to the “duopoly” of Anthropic and OpenAI.

In a recent interview on CBS’s “Face the Nation,” Delangue mentioned that Hugging Face utilized an Nvidia-modified version of a Chinese open-source model to recover from a cyberattack. He also referred to a letter jointly signed by Nvidia CEO Jensen Huang and 24 other companies, including Hugging Face, appealing to the U.S. government to promote open models rather than impose restrictions. In a CNBC interview in late July, Delangue highlighted these points once more, stressing the dangers of China “clearly dominating” the open-source AI sector.

This potential acquisition could signify a rebound for Nvidia in the cloud computing sector. About a year ago, Nvidia reportedly scaled back its own cloud service, DGX Cloud. However, according to The Information, acquiring Hugging Face, which already aids developers in operating their AI models with rented computing resources, might allow Nvidia to re-enter that market without starting from scratch.

There’s also a financial safety aspect to consider. Nvidia has committed to covering the costs associated with billions of dollars in cloud computing contracts for its clients. If those clients do not fully utilize the computing resources they’ve contracted for, Nvidia could end up with excess capacity. Buying Hugging Face would position Nvidia to offer that surplus to Hugging Face’s customers.

This deal signifies a considerable increase over Hugging Face’s previously assessed valuation. The company raised $235 million in 2023, achieving a valuation of $4.5 billion. This funding round was led by Salesforce Ventures, with participation from Alphabet’s GV, IBM Ventures, and Nvidia.

This wouldn’t be the first instance in which Hugging Face has considered Nvidia’s investment proposals. The company turned down a $500 million investment offer from Nvidia late last year, which would have set its valuation at $7 billion, according to the Financial Times. Hugging Face expressed its intention to avoid a dominant investor that could sway its decision-making.

As for why they might be more open to an offer now, one could argue that an acquisition differs from merely accepting a single major investor. The latter often entails giving up control and facing pressures to achieve rapid growth.

Furthermore, Hugging Face remains relatively modest in terms of revenue within the AI landscape. The Information reported that it recently generated around $150 million in annual revenue, a rise from approximately $100 million just two months prior.

This growth has brought the company “close to profitability,” as Delangue told TechCrunch last month. However, a valuation nearing $13 billion would represent a significant multiple for a company of this size, making it challenging to decline.

Finally, being acquired would provide Hugging Face access to Nvidia’s considerably larger financial resources just as other competitors in AI infrastructure are being absorbed by larger entities, as illustrated by Stripe’s recent acquisition of OpenRouter, a startup founded in early 2023 that aids users in selecting various AI models based on their needs and budgets.

OpenRouter had a valuation of only $1.3 billion back in May during its Series B funding round. Stripe reportedly spent over $7 billion to acquire it earlier this month.