Nvidia, to be clear, was already a shareholder in Hugging Face, having participated in a 2023 funding round that valued the startup at $4.5 billion. But Nvidia also announced a monster earnings report on Wednesday, reporting revenue of $96.2 billion in Q2 2027, up 106% from a year earlier, while guiding to $108 billion for the current quarter. Shares were up about 5% in pre-market trading on Thursday.
Assuming Hugging Face is a done deal, a $12.9 billion acquisition would be one of the biggest deals Jensen Huang’s chipmaker has ever completed. It would also be a nice full circle moment for Nvidia, which had been turned down by Hugging Face before.
What is Hugging Face?
Hugging Face has evolved drastically since its founding. It was started by three French entrepreneurs — Clément Delangue, Julien Chaumond and Thomas Wolf — back in 2016, who named it after the namesake emoji. Their very first product was a chatbot app aimed at teenagers. TechCrunch described it as an "artificial BFF" at the time.
But, as the three founders learned, building a decent chatbot meant creating their own natural-language tooling — and when Google published its open-source language model, called BERT, in late 2018, Wolf's team rebuilt it in PyTorch, an open-source deep-learning framework, and posted the code publicly in a matter of days. The reaction from others online dwarfed anything the chatbot had generated. That code became the Transformers library, and by 2019, Hugging Face had abandoned consumer apps to build infrastructure instead.
Today, Hugging Face runs the Hub, which is kind of like GitHub for machine learning and AI. Developers upload model weights (the parameters that make an AI model do what it does), datasets (the raw material models learn from) and Spaces (small hosted web apps that let anyone try a model in a browser without writing code). Industry research firm Contrary put the Hub at more than 2.4 million models and over 730,000 datasets as of January 2026, with growth accelerating since — and almost all of it is free.
You’re probably wondering how Hugging Face is makes money when so much of what it produces is free. Well, Hugging Face makes its money from enterprise subscriptions, paid GPU compute and something called “hosted inference endpoints,” which is basically a thin commercial layer sitting on top of that enormous free pile of stuff. It’s been hard to nail down just how much Hugging Face makes in revenue, but Business Insider pegged the company at roughly $100 million in annual recurring revenue, while The Information cited about $150 million annualized. At $12.9 billion, that works out to somewhere between 86 and 129 times sales. Delangue told TechCrunch's Equity podcast recently that the company is "close to profitability" and had only just started spending the money it raised three years ago.
On that topic, Hugging Face’s funding round in 2023 also got it a lot of attention. It raised $235 million at a $4.5 billion valuation, led by Salesforce Ventures with participation from Google, Amazon, IBM, Intel, AMD, Qualcomm… and, you guessed it, Nvidia. Still, though, the price was reportedly more than 100 times revenue. The current figure would nearly triple that valuation in three years.
And, as we said before, Nvidia tried to get in on the Hugging Face action earlier, but was turned down. Back in January, the Financial Times said Hugging Face had rejected a $500 million investment from the chipmaker in late 2025, which would have valued the company around $7 billion. The reason, according to TechCrunch, was that Hugging Face was trying to avoid a situation where a single investor could influence its decision-making.
Must Read
Join 250,000+ readers and get Moneywise’s best stories and exclusive interviews first — clear insights curated and delivered weekly. Subscribe now.
Why would Nvidia would pay so much for a company that doesn't make chips?
While most people know Nvidia for its silicon — its graphics chips that power computers, and its AI chips that power data centers — Nvidia’s secret sauce is actually CUDA, which is a software layer popular with developers. But now, with so many developers moving to Hugging Face, which tends to be the first stop for engineers hunting for an AI model and/or the last stop before putting a model into production, you can see why Nvidia wants a piece of the action.
