Nvidia CEO Jensen Huang announced on his blog that the company would be acquiring Hugging Face for $12.9bn.
He reassured artificial intelligence (AI) developers of open models and their users that Hugging Face would remain an open platform for the entire AI ecosystem.
“Developers will choose the models they want, the frameworks they want, the clouds and inference service providers they want and the computing platforms they want,” Huang said. He also confirmed that the company would not make Nvidia hardware a requirement for building or deploying through Hugging Face.
Commenting on the acquisition, Forrester principal analyst Naveen Chhabra said: “Nvidia gains visibility into customer’s preferences and the AI models they use. They can see which models are trending, what datasets customers are downloading and the architectures that are gaining traction weeks before they hit mainstream tech news. It turns Hugging Face into a corporate scouting ground and an unparalleled market intelligence machine.”
Chhabra said that through the acquisition of Hugging Face, Nvidia also gains data-driven foresight into developer habits, which provides a roadmap to where next-generation chip architectures need to head.
Huang claimed Hugging Face would continue to support multicloud and multiaccelerator development and deployment, enabling model developers and the users of those models to use the hardware and infrastructure that best fit their work.
Forrester’s vice-president principal analyst Jeff Pollard said: “By acquiring Hugging Face, Nvidia gains control over and visibility into a key AI supply chain layer. The company now sits closer to the entire AI supply chain: compute, models, datasets and developer workflows. That creates an opportunity to improve AI security at scale through better model transparency, provenance, vulnerability disclosure and guardrails. It also concentrates influence over a critical piece of the open AI ecosystem.”
Greater openness is one of the main focus points for Huang. In July, Nvidia – along with several other companies – announced the industry-wide Open Secure AIIiance. At the time, Nvidia described open models and open harnesses as “essential” as they democratise defensive capabilities and increase transparency for those tasked with defending IT systems.
Specifically, Nvidia sees open models as a way to enable cyber defence while complementing frontier closed models with customisable, localised controls. At the time, Nvidia said: “Open source enables massively distributed community-driven and self-controlled defence – with no single point of failure.”
In the blog, Huang said: “Open weights broaden access to AI and help to ensure that AI leadership is distributed across companies, institutions and communities.”
Announcing the Hugging Face acquisition, Huang added that open models enable startups, businesses, universities and public institutions to build on advanced capabilities without training every model from scratch: “They enable organisations to match the right model to the right job. That is how AI can advance safely, strengthen cyber security and sovereignty, accelerate innovation; and reach factories, hospitals, farms, classrooms and Main Street businesses around the world.”
Eugene Cheah, CEO and co-founder of serverless inference platform Featherless.ai, described the acquisition as a “GitHub moment”, referring to the $7.5bn acquisition of GitHub by Microsoft in 2018.
“This deal is an incredibly important milestone for the AI industry,” said Cheah. “In the same way that GitHub became a critical part of the open source ecosystem under Microsoft, Hugging Face is now ideally positioned to become its counterpart for open source AI. In short, Hugging Face is now poised to be at the centre of the nexus of AI open source development.”
