Nvidia confirms $12.9B acquisition of AI hosting platform Hugging Face
Nvidia confirmed its $12.93B acquisition of Hugging Face, gaining an open-source AI hub and model trend visibility.
CNBC reported that acquisition talks began a few weeks ago. Business Insider first wrote about the discussions last Monday, and two days later The Information reported that Nvidia had clinched the deal.
Some of the purchase price will go to chipmakers Advanced Micro Devices, Intel and Qualcomm, which backed Hugging Face’s most recent funding round. Salesforce also participated in that round and reportedly sought to buy the company before Nvidia stepped in.
Hugging Face operates a GitHub-like platform where researchers share open-source AI models and training datasets. Nvidia Chief Executive Jensen Huang wrote in a blog post that the platform hosts more than 500,000 datasets and four times as many models.
Hugging Face’s main source of revenue is cloud services for AI development teams. Its Hub service offers private versions of the hosting platform, and other services let developers run inference models on managed cloud infrastructure. Those paid products are probably not the main reason Nvidia is buying the company.
Forrester principal analyst Naveen Chhabra said the deal gives Nvidia visibility into customer preferences and the AI models they use. Nvidia can see which models are trending, which datasets customers download and which architectures are gaining traction before those trends reach mainstream tech news, he said.
Huang wrote that Hugging Face will continue to support AI projects that can run across multiple chips and clouds, and that Nvidia will not require users to deploy workloads on its silicon. Chhabra added that Nvidia gains data-driven foresight into what next-generation chip architectures to build, and that the company is securing the software layer because hardware-only companies tend to be commoditized over time.
The acquisition is Nvidia’s second-largest startup deal, after its $20 billion licensing agreement with Grok Inc. last year. That transaction gave Nvidia access to an inference-optimized chip architecture and related talent, which it used to develop the Groq 3 LPU processor.
Last week, Hugging Face introduced a rollerblade-equipped robotic duck that can pick up items with its beak; sales of the device topped $5 million in five days. The company also has worked with more than 1,000 researchers to develop its own large language model.