Nvidia has agreed to acquire Hugging Face, the open-source artificial intelligence platform, for $12.93 billion in one of the biggest takeovers yet of a company built around openly available AI models.
The deal, announced by Nvidia chief executive Jensen Huang on Thursday, will bring the platform’s extensive developer community and model library under the control of the world’s dominant AI chipmaker.
Hugging Face was founded in New York in 2016 by French entrepreneurs Clément Delangue, Julien Chaumond and Thomas Wolf. Its name was inspired by the hugging face emoji, but the company has since grown into a central hub for developers building, sharing and deploying AI systems.
Nvidia said more than 18 million developers, researchers and creators use the platform to share more than three million models, 500,000 datasets and one million applications. More than 200,000 companies use Hugging Face to discover, assess, customise and deploy AI tools.
The transaction includes approximately $11.9 billion to be paid to Hugging Face shareholders, subject to adjustments, alongside an employee retention programme worth up to about $1 billion, according to a regulatory filing. Nvidia expects the acquisition to close in the first half of 2027, pending customary conditions and regulatory approvals.
Huang said Hugging Face would remain “an open platform for the entire AI ecosystem”. Developers, he added, would continue to choose which models, frameworks, cloud providers, inference services and computing platforms they used.
“Nvidia compute will not be required to build on or deploy through Hugging Face,” the company said.
The assurance will be closely watched across the AI industry. Nvidia’s graphics processors have become the default infrastructure for training and running advanced models, while Hugging Face has positioned itself as a relatively neutral marketplace and collaboration platform for open-weight AI.
The acquisition gives Nvidia a direct stake in a part of the ecosystem that could become increasingly important as major customers such as OpenAI, Microsoft, Amazon and Meta develop their own chips and seek to reduce their reliance on Nvidia hardware.
Why Nvidia is buying Hugging Face
Open-weight models can be downloaded, modified and run by organisations on their own systems, offering greater control and customisation than closed AI services. Nvidia’s ownership of Hugging Face could help it promote that model of development while creating new routes for businesses to use its computing infrastructure.
Nvidia has already been a significant contributor to the platform, with more than 500 models and 250 open datasets listed on Hugging Face. The acquisition will give it greater influence over the place where many developers discover, test and distribute AI models.
Hugging Face had previously resisted being absorbed by Nvidia. The chipmaker took part in the company’s 2023 funding round, which valued the start-up at $4.5 billion, and was reported to have offered $500 million last year at a valuation of $7 billion.
That proposal was understood to have raised concerns among the founders because it would have made Nvidia the largest minority shareholder and threatened the company’s independence.
Delangue did not confirm the reported offer during a briefing with reporters, saying publicly reported accounts of Hugging Face’s fundraising could be “quite far from reality”. He said the company had received several acquisition approaches over the years but had concluded that Nvidia was now the right owner.
“I think this summer the planets aligned,” he said, adding that Hugging Face had become increasingly convinced Nvidia would be “the perfect home” for the company and the wider open AI community.
Wolf, one of the company’s co-founders, said the team had not expected the field to expand so dramatically when it began as a small start-up in 2016.
For Hugging Face, Nvidia’s financial backing could accelerate its ambition to reach 100 million AI builders in the coming years. Delangue said the company believed Nvidia’s support would give it a better chance of reaching that target more quickly.
But the deal has also raised questions about who benefits financially from the value created on the platform. Developers who build and upload models have generally not been paid for downloads or use of their work.
Eric Hartford, the creator of the open-weight Dolphin model, said Hugging Face had never found a way to monetise such creations. “If I got a dollar for every download of Dolphin I’d be rich,” he said.
Critics also fear that Nvidia’s ownership could influence the technical direction of Hugging Face, even if the platform remains formally open. Model creators may be encouraged to optimise their work for Nvidia hardware, while competing chips and computing systems could become less prominent.
Hartford said Hugging Face’s Transformers library already played a key role in model releases and warned that Nvidia ownership could create incentives to favour the company’s products. Nvidia has rejected that concern, insisting that developers will remain free to use rival hardware and services.
The takeover marks a major test for the credibility of open AI. It could provide Hugging Face with the resources to broaden access to powerful models, while leaving developers to judge whether the platform can retain the neutrality that helped make it so influential.
