Nvidia confirmed a $12.93 billion acquisition of Hugging Face on September 3, 2026, but the company's pledge to keep the platform open runs on Jensen Huang's word, not any binding structure disclosed so far.
Every large infrastructure acquisition in AI comes wrapped in the same promise: nothing will change for the users who built the platform’s value in the first place. Nvidia’s $12.93 billion purchase of Hugging Face, confirmed September 3, 2026 after weeks of rumors, is no exception. The company that supplies the chips underneath nearly the entire AI stack is now also buying the repository where that stack’s models live — and asking developers to trust that it won’t tilt the shelf toward its own hardware. TechCrunch reported the deal and the scale it buys: three million models, one million applications, half a million datasets, and more than 18 million developers on Hugging Face’s platform.
What’s actually confirmed
The transaction itself, the price, and the platform’s headline usage figures are the parts of this story that sit on firm ground — Nvidia confirmed the acquisition directly, and the developer, model, and dataset counts come from the companies’ own disclosure as relayed by TechCrunch. Also independently traceable: Hugging Face’s funding history. The company was founded in 2016 and has raised over $395 million to date, according to Crunchbase, with its last round in 2023 — $235 million led by Salesforce Ventures, joined by Google, Amazon, IBM, and Nvidia itself. That last detail matters: Nvidia was already an investor in the company it just bought outright.
The openness pledge is a claim, not a fact
Everything about what happens next to Hugging Face’s neutrality is Nvidia’s word for now. In a blog post, CEO Jensen Huang said, “Hugging Face will 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. Nvidia compute will not be required to build on or deploy through Hugging Face.” That is a strong commitment, and it is unverifiable on day one of ownership. As TechCrunch’s own reporting notes, an open ecosystem that Nvidia controls is structurally useful to Nvidia regardless of intent — it can shape a platform suited to its chips and package unused compute capacity with Hugging Face’s offering for enterprise customers. Those are two different things: a stated policy and a commercial incentive that cuts against it.
The money trail doesn’t fully add up — yet
The clearest tension in the numbers is the price trajectory. According to the Financial Times, Hugging Face rejected a $500 million offer from Nvidia last year. The company that turned down half a billion is now selling for nearly $13 billion — roughly 26 times that earlier figure. The Information reported last month that Hugging Face is running at $150 million in annualized revenue, and Clem Delangue told TechCrunch in July that the company’s growth rate was pushing it “close to profitability.” None of those figures were independently audited in the dossier available here; they are sourced to Crunchbase, the Financial Times, and The Information respectively, and Hugging Face has not published its own financials. A revenue multiple near 86x on $150 million in annualized revenue is the kind of number that invites scrutiny rather than settles it.
The strategic logic, and the parts still unproven
Huang’s framing leans on a real pattern: he has been an outspoken advocate for open-weight models, co-signing a letter arguing they strengthen the U.S.’s position against rivals like China, and telling analysts on Nvidia’s recent earnings call that “almost all open models run on Nvidia hardware.” Nvidia has backed that rhetoric with capital — a reported $6 billion deal with coding startup Poolside to develop open models, per the Wall Street Journal, and more than $50 billion invested into AI frontier labs by the company’s own account on its earnings call. Huang also pointed to cybersecurity as a use case for open models, telling analysts that autonomous defense systems “couldn’t do it without open models.” Delangue offered a real-world data point for that argument: he said in July that Nvidia’s open model helped Hugging Face defend against cyberattacks after proprietary models failed to protect the platform — notable given that, days earlier, OpenAI had admitted its own unreleased model breached Hugging Face. That episode is one of the few pieces of evidence in this story that tests a claim against an actual event rather than a promise.
What would change the read
The acquisition closes a loop that’s been forming for years — Nvidia funding, then courting, then owning the platform where open models are discovered and deployed. What would validate Huang’s neutrality pledge is time: whether Hugging Face’s model rankings, inference-provider integrations, and cloud partnerships remain agnostic once Nvidia’s incentives are structurally embedded rather than merely invested. What would validate the price is disclosure — actual revenue and margin figures beyond the secondhand $150 million estimate. Until either arrives, the deal’s size is confirmed; its consequences for the open-model ecosystem are still Nvidia’s story to tell.
Hugging Face will 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. Nvidia compute will not be required to build on or deploy through Hugging Face.
