Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Nvidia has agreed to buy AI model platform Hugging Face for $13bn, in the latest step by the $5.4tn chip giant to use its financial might to accelerate the technology’s boom while exerting greater control over the industry.
Hugging Face, which only last year turned down a large investment from Nvidia at a $7bn valuation to maintain its independence, serves as a repository for millions of models and data sets and has become a champion of “open” AI systems.
Nvidia said the goal of the deal was to speed up the spread of open models. Unlike proprietary models from labs such as OpenAI and Anthropic, the design of open-weight models is public and users can download, customise and run them on their own hardware.
“Open weights enable start-ups, established businesses, universities and public institutions to build on advanced capabilities without training every model from scratch or paying frontier-model prices for every task,” Nvidia chief executive Jensen Huang said in a statement.
“That is how AI can scale sustainably into billions of everyday tasks across factories, hospitals, farms, classrooms and Main Street businesses,” he added.
Nvidia, the world’s most valuable company, is the largest designer of the advanced chips used to train and run AI models. It has committed hundreds of billions of dollars in start-up investments, loan backstops and financial guarantees in the name of expanding the ecosystem, thereby generating more demand for its products.
Thursday’s deal is the largest outright acquisition of a company by Nvidia, far eclipsing its $6.9bn purchase of networking technology company Mellanox in 2020 — a deal that laid the foundations for Nvidia’s move beyond chips into offering full data centre infrastructure.
Nvidia hopes to close the Hugging Face takeover by 2027, although the transaction is likely to face scrutiny from competition regulators.
It will give Nvidia, which has developed its own family of models, control of one of the key distribution channels that influences how widely certain AI applications are adopted around the world.
“We will have to get through all the regulatory review. But we think overwhelmingly they’re going to see this as really a positive outcome,” said Justin Boitano, vice-president of enterprise AI at Nvidia.
Nvidia pledged that Hugging Face “will remain an open platform for the entire AI ecosystem”, with the more than 18mn developers who use the platform able to choose which models, cloud providers and chips they use.
The platform, named for the “hugging face” emoji, hosts 3mn primarily “open” AI models, about 500,000 datasets and 1mn AI applications. More than 200,000 companies use it to discover AI features, Nvidia said.
The proliferation of open-weight models would also create new channels of demand for Nvidia’s chip business, easing its reliance on a handful of customers, including OpenAI and Anthropic, who are also developing their own AI processors. Huang has thrown his company’s financial muscle behind groups developing open models, such as Reflection AI, and the infrastructure underpinning them, backing so-called neoclouds including CoreWeave and Nebius.
New York-based Hugging Face was recently at the centre of a high-profile hacking incident when OpenAI models escaped human control during testing and broke into the platform.
The 10-year-old company, which has taken venture capital funding from Nvidia in the past alongside Google, Amazon, Intel and others, makes money from premium subscriptions and enhanced services for corporate users. Nvidia plans to maintain the Hugging Face brand.
It turned down a $500mn investment from Nvidia late last year, in part to avoid having a single dominant investor, the FT reported.
Nvidia has recently stepped up its advocacy for open models amid a political tussle in the US over whether Chinese developers of increasingly advanced open systems should be sanctioned in relation to claims they have stolen technology from American labs.
Huang put his name to a letter in July calling for the US to support the open models. “AI leadership will be judged not by one frontier AI model, but by whether the United States builds a strong, open ecosystem that diffuses into every sector,” the letter said.
Additional reporting by George Hammond
