Nvidia in Advanced Talks to Acquire Hugging Face for $12.9 Billion Amidst Strategic Open-Source AI Push

Nvidia has reportedly agreed to acquire Hugging Face, a prominent open-source AI model repository, for an estimated $12.9 billion. The Information first broke the news on Wednesday night, citing a source familiar with the ongoing discussions. However, Business Insider, which initially reported over the weekend that Hugging Face was exploring takeover interest, cautioned on Wednesday night that a signed agreement had not yet materialized, and the talks, which could value the company at over $13 billion, still had the potential to fall apart.

As of the reporting, neither Nvidia nor Hugging Face had responded to inquiries from TechCrunch regarding the potential acquisition. Nvidia’s silence is particularly notable, given its historical tendency to quickly address reports it deems inaccurate, suggesting the information might hold substantial weight.

The acquisition, if finalized, would represent a significant strategic move for Nvidia. Hugging Face, established in 2016, has grown into one of the most widely used platforms where developers actively share, discover, and download open-source artificial intelligence models. Integrating Hugging Face into its ecosystem would provide Nvidia with a substantial and immediate foothold in the burgeoning world of open-source AI, a sector that is currently experiencing rapid growth as open-source developers strive to match and even surpass the capabilities of proprietary AI systems offered by major players like Anthropic and OpenAI.

Nvidia’s motivation for such a substantial investment appears to be multifaceted, primarily centered on safeguarding and expanding its dominant position in the highly competitive AI chip market. From an external perspective, Nvidia’s supremacy in AI hardware, despite its aggressive chip-release schedule, appears increasingly vulnerable. Nearly all of the leading closed-source AI laboratories, including industry giants such as OpenAI, Google, Amazon, and Anthropic, are actively developing their own custom AI chips. This internal development aims to reduce their dependence on Nvidia’s hardware, posing a long-term threat to the chipmaker’s market share.

In this context, fostering a vibrant and robust ecosystem of open-source AI models becomes a critical strategic imperative for Nvidia. By supporting and nurturing open-source alternatives, Nvidia can offer customers more choices beyond the closed, proprietary systems offered by its competitors. This, in turn, helps to ensure that a larger segment of the AI market remains reliant on Nvidia’s foundational hardware infrastructure. This strategic alignment is consistent with Nvidia’s existing investments; the company has already committed tens of billions of dollars towards developing its own open-source AI models, underscoring its belief in the long-term value of this approach.

The prospect of Hugging Face relinquishing its independence is not entirely surprising, particularly considering the recent public alignment of its CEO, Clem Delangue, with Nvidia’s advocacy for open-source AI. This alignment has unfolded amidst a broader, months-long debate, with Washington officials reportedly weighing potential restrictions on open-weight models. The debate intensified due to the emergence of Chinese laboratories, such as Moonshot AI, which have released systems like its Kimi K3 model that have demonstrated performance benchmarks comparable to leading U.S. models, often at a significantly lower operational cost. This development has fueled concerns in Washington regarding competitive parity and national security. Critics of the closed-lab model, including White House advisor David Sacks, have suggested that these fears might be amplified by the "duopoly" of Anthropic and OpenAI, potentially to their advantage.

Clem Delangue has been a vocal proponent of open models. In an appearance on CBS’s "Face the Nation" earlier this month, Delangue highlighted how Hugging Face utilized an Nvidia-modified version of a Chinese open-source model to defend itself against a cyberattack. He also referenced a recent letter, co-signed by Nvidia CEO Jensen Huang, Delangue himself, and 24 other companies, which collectively urged the U.S. government to actively support open models rather than imposing restrictions. In a separate interview with CNBC in late July, Delangue reiterated these points, again citing the influential letter and issuing a stark warning that China is "clearly dominating" the open-source AI landscape, emphasizing the urgency for U.S. support.

Beyond protecting its chip dominance, the potential acquisition also signals a strategic resurgence for Nvidia in the realm of cloud computing. Approximately a year ago, Nvidia reportedly scaled back its own cloud business, known as DGX Cloud, refocusing its cloud efforts internally. However, as reported by The Information, acquiring Hugging Face, a platform that already facilitates developers in running their AI models using rented computing power, could provide Nvidia with a direct and established pathway back into the cloud market without the necessity of building an infrastructure from the ground up.

Furthermore, a significant financial safety net appears to be at play for Nvidia. The company has made substantial commitments, promising to help cover tens of billions of dollars in cloud computing deals for its customers. Should these customers ultimately not utilize the full computing power they initially signed up for, Nvidia could find itself burdened with significant unused capacity. Owning Hugging Face would offer Nvidia a practical solution: the ability to efficiently sell this potentially unused capacity to Hugging Face’s extensive customer base, thereby mitigating financial risk and optimizing resource utilization.

The reported acquisition price of nearly $13 billion marks a colossal increase from Hugging Face’s last publicly known valuation. In 2023, the company successfully raised $235 million in a funding round that valued it at $4.5 billion. That particular round was spearheaded by Salesforce Ventures, with additional investments from notable entities including Alphabet’s GV, IBM Ventures, and even Nvidia itself, among other contributors.

This would not be Hugging Face’s first interaction with an Nvidia offer. The Financial Times previously reported that Hugging Face declined a $500 million investment offer from Nvidia late last year, which would have valued the company at $7 billion. At the time, Hugging Face reportedly expressed a preference against having a dominant investor that could potentially influence its strategic decisions.

Regarding the shift in stance, one could argue that a complete buyout presents a fundamentally different scenario compared to accepting a significant investment from a single, powerful backer. The latter often entails ceding a degree of control and facing continuous pressure for accelerated growth.

Despite its influential role in the AI developer community, Hugging Face remains a comparatively smaller business in terms of revenue within the broader AI industry. The Information reported that the company was recently generating approximately $150 million annually in revenue, a notable increase from roughly $100 million just two months prior. This growth trajectory has brought the company "close to profitability," as CEO Clem Delangue informed TechCrunch last month. Nevertheless, a valuation approaching $13 billion represents a massive multiple for a company of this revenue size, making such an offer exceedingly difficult to decline.

Finally, the acquisition would grant Hugging Face access to Nvidia’s considerably deeper financial resources, a crucial advantage in a rapidly consolidating market. The AI infrastructure sector is witnessing increasing integration, as exemplified by Stripe’s recent acquisition of OpenRouter. OpenRouter, a startup founded in early 2023, specializes in assisting customers in selecting various AI models tailored to their specific needs and budgetary constraints. OpenRouter was valued at just $1.3 billion during its Series B round in May, yet Stripe reportedly paid over $7 billion to acquire it earlier this month, highlighting the premium placed on strategic AI infrastructure assets.