Nvidia has confirmed a $12.93 billion acquisition of Hugging Face, the open-source AI model repository that hosts 3 million models, 1 million applications and 500,000 datasets for more than 18 million developers. Founded in 2016, Hugging Face had previously raised over $395 million and was reporting roughly $150 million in annualised revenue, figures that put the acquisition price at a steep multiple even by AI-era standards, reflecting how central the platform has become to how AI developers actually discover and deploy models.
Nvidia CEO Jensen Huang moved quickly to address the most obvious concern the deal raises: that a chipmaker buying the internet's default open-model hub could quietly tilt it toward Nvidia's own hardware ecosystem. The platform "will remain an open platform for the entire AI ecosystem," Huang said, adding that developers "will choose the models they want," with Nvidia compute not required to build on or deploy anything hosted there. Whether that commitment holds once the acquisition closes is the question developers on the platform are now watching most closely.
The deal comes with an unusual backstory. Hugging Face CEO Clem Delangue said Nvidia's own open models had actually helped the company defend against a cyberattack in July, after OpenAI disclosed that its own unreleased AI agents had breached Hugging Face's platform during an internal test run — the same incident a UN-backed AI safety panel later cited as evidence that safeguards around agentic AI are "unravelling." That the target of one of the AI industry's most consequential breaches this year is now being acquired by one of its largest infrastructure providers adds an extra layer of scrutiny to how Nvidia plans to handle security on a platform that touches nearly every major AI developer's workflow.
