wouldit.work

AI tools

Hugging Face: a $12.93 billion Nvidia deal after a $4.5 billion Series D

Clément Delangue, Julien Chaumond, and Thomas Wolf started Hugging Face as a chatbot in 2016, then built a hub for open AI models. In 2023 it raised a $235 million Series D at a $4.5 billion valuation. In September 2026 Nvidia agreed to acquire the company for $12.93 billion.

Founder
Clément Delangue, Julien Chaumond, and Thomas Wolf
Launched
2016
Model
Paid inference, training, and enterprise hub features on top of a free model hub

Published · 3 min read · 5 sources

Key takeaways

  • Delangue, Chaumond, and Wolf launched Hugging Face in 2016 as a chatbot for teenagers, then pivoted after open-sourcing the model behind it.
  • In August 2023 the company raised a $235 million Series D at a $4.5 billion valuation, with Google, Amazon, Nvidia, and Salesforce among the investors.
  • On 3 September 2026 Nvidia CEO Jensen Huang wrote that Nvidia had agreed to acquire Hugging Face for $12,930,300,000 and that the hub would remain an open, multi-cloud platform.

What it is

Hugging Face is a platform where developers share, train, and deploy machine-learning models, datasets, and apps. Co-founder and CEO Clément Delangue launched the Brooklyn-based company in 2016 with Julien Chaumond and Thomas Wolf. TechCrunch wrote that they "originally built a chatbot app targeted at teenagers," then "pivoted to focus on creating a platform for creating, testing and deploying machine learning" after open-sourcing the algorithm behind the app (TechCrunch).

Timeline

  • 2016: The company launched as a consumer chatbot, then shifted to developer tools and a model hub (TechCrunch).
  • August 24, 2023: Hugging Face raised $235 million in a Series D that valued the company at $4.5 billion. TechCrunch, first pointing to The Information and then to a post by Salesforce CEO Marc Benioff, listed participation from Google, Amazon, Nvidia, Intel, AMD, Qualcomm, IBM, Salesforce, and Sound Ventures. Delangue told the publication, "AI is the new way of building all software." The company claimed 10,000 customers, "more than 50,000 organizations on the platform," a hub with "over 1 million repositories," and 170 employees, and said it had raised a total of $395.2 million (all company-reported) (TechCrunch).
  • August 26, 2026: TechCrunch reported that talks about an Nvidia purchase were underway, citing The Information on a $12.9 billion figure and Business Insider on the possibility that no agreement had been signed yet (TechCrunch).
  • September 3, 2026: Nvidia CEO Jensen Huang wrote that "NVIDIA has agreed to acquire Hugging Face for $12,930,300,000." He said "more than 18 million developers, researchers and creators" used the platform to share "more than 3 million models, 500,000 datasets and 1 million applications," and that "more than 200,000 companies" used it (figures stated by Nvidia). "Hugging Face will remain an open platform for the entire AI ecosystem," he wrote. "NVIDIA compute will not be required to build on or deploy through Hugging Face." Delangue, quoted by TechCrunch from a post on X, said the company went to Huang because scaling open models needed "more compute, more support, more collaboration, and more visibility" (Nvidia; TechCrunch).

Why it worked

  • The hub became default infrastructure. By 2023 the company was already selling inference, training, and an enterprise hub on top of a public repository that labs and clouds wanted to be on (TechCrunch).
  • Chip and cloud vendors paid to be close to the developers. The Series D roster included Nvidia, Salesforce, Google, and Amazon, which were also partners for compute and hosting (TechCrunch).
  • The buyer wanted the ecosystem, not just the revenue. Huang's post stressed that the hub would stay multi-cloud and multi-accelerator, a promise aimed at the community that made the asset valuable (Nvidia).

What to learn

  1. A failed consumer app can still leave you with the library people actually wanted (TechCrunch).
  2. If your marketplace is where models live, incumbents may invest first and bid later (TechCrunch; Nvidia).
  3. "We will remain open" is a statement about the present deal, not a guarantee after closing. Huang's wording is worth keeping, and checking, after the acquisition completes (Nvidia).
  4. Strategic buyers can pay a price that dwarfs the last venture mark: from $4.5 billion in 2023 to $12.93 billion in the Nvidia agreement.

Risks and what didn't work

  • The deal, as announced, was an agreement to acquire, not proof that it had already closed. This article uses Huang's "agreed to acquire" language (Nvidia).
  • Developer, model, and company counts are stated by Nvidia or by Hugging Face, not independently audited here (Nvidia; TechCrunch).
  • TechCrunch, citing The Information, wrote that the company was 'recently generating about $150 million a year in revenue.' That figure is second-hand and was not confirmed in Huang's post (TechCrunch).
  • Openness after a chip-maker acquisition is a standing tension, however carefully the announcement is worded (Nvidia).

FAQ

Short answers drawn from the sections above.

What is Hugging Face?

Hugging Face is a platform where developers share, train, and deploy machine-learning models, datasets, and apps. Co-founder and CEO Clément Delangue launched the Brooklyn-based company in 2016 with Julien Chaumond and Thomas Wolf. TechCrunch wrote that they "originally built a chatbot app targeted at teenagers," then "pivoted to focus on creating a platform for creating, testing and deploying machine learning" after open-sourcing the algorithm behind the app (TechCrunch).

Why did Hugging Face work?

  • The hub became default infrastructure. By 2023 the company was already selling inference, training, and an enterprise hub on top of a public repository that labs and clouds wanted to be on (TechCrunch).
  • Chip and cloud vendors paid to be close to the developers. The Series D roster included Nvidia, Salesforce, Google, and Amazon, which were also partners for compute and hosting (TechCrunch).
  • The buyer wanted the ecosystem, not just the revenue. Huang's post stressed that the hub would stay multi-cloud and multi-accelerator, a promise aimed at the community that made the asset valuable (Nvidia).

What can indie builders learn from Hugging Face?

  1. A failed consumer app can still leave you with the library people actually wanted (TechCrunch).
  2. If your marketplace is where models live, incumbents may invest first and bid later (TechCrunch; Nvidia).
  3. "We will remain open" is a statement about the present deal, not a guarantee after closing. Huang's wording is worth keeping, and checking, after the acquisition completes (Nvidia).
  4. Strategic buyers can pay a price that dwarfs the last venture mark: from $4.5 billion in 2023 to $12.93 billion in the Nvidia agreement.

What didn't work for Hugging Face, and what are the risks?

  • The deal, as announced, was an agreement to acquire, not proof that it had already closed. This article uses Huang's "agreed to acquire" language (Nvidia).
  • Developer, model, and company counts are stated by Nvidia or by Hugging Face, not independently audited here (Nvidia; TechCrunch).
  • TechCrunch, citing The Information, wrote that the company was 'recently generating about $150 million a year in revenue.' That figure is second-hand and was not confirmed in Huang's post (TechCrunch).
  • Openness after a chip-maker acquisition is a standing tension, however carefully the announcement is worded (Nvidia).

Sources

Every link cited above, in order of first use.

  1. 1.Hugging Face (huggingface.co)
  2. 2.TechCrunch (techcrunch.com)
  3. 3.$12.9 billion (techcrunch.com)
  4. 4.Nvidia (blogs.nvidia.com)
  5. 5.TechCrunch (techcrunch.com)

Related case studies

More in AI tools →