Nvidia has agreed to acquire Hugging Face in a deal valued at approximately $12.93 billion, a transaction that would bring one of the industry’s most important open-model platforms under Nvidia ownership.
The companies announced the deal Thursday, Sept. 3, saying Nvidia will provide the infrastructure, engineering resources and global reach to scale Hugging Face while preserving the openness and hardware neutrality that helped make the platform a central hub for AI developers.
Hugging Face currently serves more than 18 million developers, researchers and creators and hosts more than 3 million models, 500,000 datasets and 1 million applications. More than 200,000 companies use the platform to discover, evaluate, customize and deploy AI, according to Nvidia.
Nvidia makes a $12.9 billion bet on open models
Nvidia CEO Jensen Huang framed the acquisition as an investment in expanding access to open models rather than in pulling Hugging Face into Nvidia’s proprietary ecosystem.
The company said Hugging Face will continue to support models from across the industry, as well as multi-cloud and multi-accelerator deployments. Nvidia compute will not be required for developers building or deploying through Hugging Face.
“I am honored that Clem came to me as he considered the next chapter of Hugging Face and believed NVIDIA would be a great home for the company, its community and the future of open models,” Huang’s statement reads in part.
That commitment is strategically important for both companies.
Nvidia VP of Enterprise AI Justin Boitano said during a Q&A following the announcement that Nvidia sees both closed and open models as necessary to the broader AI market. Open models, he argued, give businesses greater ability to develop domain-specific intelligence while maintaining control of their data and competitive differentiation.
For Nvidia, more development around open models still creates an economic opportunity.
“NVIDIA will benefit through the training that’s done and the inference that’s done on our hardware,” Boitano said, as open models become more widely deployed.
That illustrates the strategic logic behind the deal. Nvidia does not necessarily need Hugging Face to require Nvidia hardware. If an expanding open-model ecosystem increases overall AI development, training and inference workloads, the company remains positioned to capture a substantial share of the resulting infrastructure demand.
Hugging Face accepts deal after reported Nvidia investment talks
The deal also creates an unavoidable question about how Hugging Face’s view of independence has evolved.
The Financial Times reported earlier this year that Hugging Face rejected a $500 million investment from Nvidia that would have valued the company at $7 billion, with the startup reportedly wary of allowing a dominant investor to influence its decisions. That account was subsequently cited by TechCrunch and others as acquisition talks emerged in August.
Delangue challenged that framing Thursday, saying Hugging Face does not typically comment on fundraising conversations that fail to materialize and describing much of the previous reporting as “quite far from the reality.”
Rather than portraying the acquisition as a reversal on independence, he said Hugging Face had turned down numerous investment and acquisition approaches over the years before concluding this summer that Nvidia had become the right long-term home for the platform.
Still, the contrast is likely to follow the deal through regulatory review and into its integration.
Hugging Face says neutrality will remain central
For Hugging Face, the bigger challenge may be convincing its community that Nvidia ownership will not compromise the neutrality that made the platform valuable in the first place.
Its next target is significant: Delangue said Hugging Face wants to grow from 18 million AI builders today to 100 million within the next few years.
Both executives repeatedly emphasized that doing so requires maintaining developer trust.
Boitano said Hugging Face users must continue to access models and datasets on the infrastructure of their choice, including other cloud and hardware providers. Delangue went further, pointing to the open-source nature of much of Hugging Face’s technology as a structural safeguard.
Those promises will likely become one of the most closely watched aspects of the acquisition. Hugging Face has become valuable precisely because developers can use it across competing model providers, clouds and hardware architectures. Any perception that the platform increasingly favors Nvidia technology could undermine the community Nvidia is paying billions to acquire.
Deal could reshape the open-source AI power structure
The acquisition also highlights a broader shift in competition across the AI market.
Nvidia has built its dominance primarily through compute, while companies including OpenAI, Anthropic and Google have built increasingly powerful proprietary model platforms. Hugging Face occupies a different position, providing infrastructure for a decentralized ecosystem of open and open-weight models that businesses can modify and deploy themselves.
Delangue argued that the open ecosystem provides a counterweight to the increasing concentration of AI capabilities behind proprietary APIs.
The irony is that one of the most prominent platforms supporting that decentralization will now belong to one of the most powerful companies in AI.
For cloud providers, infrastructure partners and enterprise AI vendors, the practical question is whether Hugging Face remains as neutral under Nvidia ownership as it is today. Nvidia has formally committed to supporting competing clouds and silicon platforms, but partners will be watching whether future integrations and product development increasingly favor Nvidia’s stack.
Nvidia and Hugging Face contend that Nvidia’s resources will make open AI more distributed, not less. That makes the success of Nvidia’s $12.93 billion bet dependent on more than Hugging Face’s technology or user count. Nvidia is also acquiring the trust of an open-source community — an asset that could prove considerably harder to own.