Hugging Face Company Profile

The CODEW · Company Profile · Company Intelligence · Enterprise AI Intelligence

Last updated: October 15, 2026 | Status: Acquired by Nvidia (September 2026) | Acquisition price: $12.93 billion

Company overview

Hugging Face is the leading platform for open-source artificial intelligence, often described as the "GitHub of machine learning." Developers and companies use its hub to share, download and run millions of open AI models, datasets and applications, and roughly half of the Fortune 500 use the platform. The company makes money by selling paid subscriptions, enterprise hub plans, inference services and compute infrastructure to the small share of users who need advanced features, while keeping the platform free for everyone else.

The platform hosts more than 3 million models, 500,000 datasets and 1 million applications, and counts 18 million developers. More than 200,000 companies use it to discover AI features. In September 2026, Nvidia acquired Hugging Face for $12.93 billion, valuing the company at roughly 86 times its $150 million in annualized revenue and nearly triple the $4.5 billion valuation it fetched in 2023. The deal was by far Nvidia's largest acquisition, far eclipsing its $6.9 billion purchase of Mellanox in 2020.

Hugging Face began in 2016 as a chatbot app for teenagers before pivoting to open-source NLP tools in 2018. It has since become the default distribution layer for open-source AI, hosting models from Meta (Llama), Alibaba (Qwen), Mistral, and Nvidia itself. The company also builds affordable open-source robots such as Reachy Mini through its acquisition of Pollen Robotics.

Key facts

Founded	2016
Founders	Clément Delangue, Julien Chaumond, Thomas Wolf
Headquarters	New York, NY (originally Paris)
CEO	Clément Delangue (co-founder)
CTO	Julien Chaumond (co-founder)
CSO	Thomas Wolf (co-founder)
Status	Acquired by Nvidia (September 2026)
Acquisition price	$12.93 billion
Revenue (run rate)	$150 million
Total funding	$395.2 million
Employees	~250–769 (sources vary)
Community	18 million developers
Enterprise customers	200,000+ companies
Content	3 million+ models, 500,000 datasets, 1 million+ applications
Paying customers	2,000+ (enterprise)

Business model

Hugging Face operates a freemium platform model. The core Hub is free and open to all, driving massive adoption and network effects. Monetization occurs through three product tiers layered on top of the free hub:

Inference Endpoints (roughly 40% of revenue): Customers pick any model from the Hub (or upload their own), select a GPU instance type (starting at $0.60/hour for a T4, up to several dollars per hour for A100s and H100s), and Hugging Face deploys it as a production-ready API with autoscaling, security, and monitoring included. It's model deployment as a managed service — the customer doesn't touch infrastructure.
Enterprise Hub (roughly 35% of revenue): At $20 per user per month, organizations like Google, Bloomberg, and Grammarly pay for private model repositories, single sign-on (SSO via SAML/OIDC), granular access controls per repo, audit logs, and priority support. Enterprise Hub also includes Inference API access with higher rate limits and dedicated endpoints — making it a bundled platform deal rather than a single feature.
Pro subscriptions (roughly 15% of revenue): At $9 per month, targeting individual power users. Pro unlocks early access to new features, higher rate limits on the free Inference API, persistent Spaces (the free tier sleeps inactive Spaces after 48 hours), and access to ZeroGPU (shared GPU instances for running models in Spaces).

The remaining ~10% comes from other services. The company's revenues come mainly from data storage services, renting out computing power to run AI models, and AI management tools. 97% of users pay nothing, and the platform remains free for the vast majority of developers.

Products and technology

Area	Status
Model Hub	3 million+ open AI models; the default distribution channel for open-source AI. Hosts Meta's Llama, Alibaba's Qwen, Mistral's open models, and Nvidia's own 500+ models.
Datasets	500,000+ datasets for training and evaluation
Spaces	1 million+ interactive AI applications; free tier sleeps inactive Spaces after 48 hours
Enterprise Hub	Paid tier with private model repositories, SSO (SAML/OIDC), granular access controls, audit logs, and priority support. $20/user/month.
Inference Endpoints	Managed inference for models hosted on the platform; production-ready API with autoscaling, security, and monitoring. Largest revenue line (~40%).
Pro Subscriptions	$9/month for individual power users; early access to features, higher rate limits, persistent Spaces, and ZeroGPU access
Transformers Library	The de facto standard Python library for state-of-the-art NLP and multimodal models; maintained by Hugging Face
Pollen Robotics / Reachy Mini	Affordable open-source robots. A rollerblade-equipped robotic duck that can pick up items with its beak sold $5 million in five days.
Cloud partnerships	Partners with AWS, Google Cloud, and Microsoft Azure to support open-source AI development across global developer communities

Financial information

Metric	Value
Annualized revenue (run rate)	~$150 million
Revenue growth	50% increase from ~$100 million in June 2026
Acquisition price	$12.93 billion
Acquisition multiple	~86x revenue
2023 valuation (Series D)	$4.5 billion (led by Salesforce Ventures)
Total funding raised	$395.2 million across 6 rounds
Revenue multiple (2023)	~64x on $70 million ARR
Founders' net worth (each)	~$1.8 billion (after Nvidia deal)
Investors in 2023 round	Salesforce Ventures (lead), Nvidia, Google, Amazon, AMD, Intel, Qualcomm, Sound Ventures, Lux Capital, Kevin Durant, and others

Leadership

Clément Delangue	Co-founder and CEO. Age 38. Background in entrepreneurship and education from Stanford University. Previously founded Moodstocks (acquired by Google). The founding team will stay on after the Nvidia acquisition.
Julien Chaumond	Co-founder and CTO. Age 42. Led development of the Transformers Library and initiatives including the BigScience project. Educated at École Polytechnique and Télécom Paris.
Thomas Wolf	Co-founder and Chief Science Officer. Age 41. Physicist turned patent lawyer before joining Hugging Face in 2017. Has been discussing the robotics direction, including the $399 open-source robot duck.

Competitors

GitHub (Microsoft): The dominant code-sharing platform; GitHub Models and Copilot compete in AI developer tooling. GitHub has the strongest developer ecosystem with high AI-assisted programming integration.
ModelScope (Alibaba) and MoArk: China's leading open-source AI model platforms, offering domestic open-source models to run natively on Chinese chips. Both provide services similar to Hugging Face, including model testing and customization.
AWS Bedrock, Google Vertex AI, Azure AI: Cloud providers offer competing model hosting and inference services, though none match Hugging Face's community scale.
OpenAI, Anthropic, Google: Closed-source AI labs that compete for developer mindshare with proprietary models.
TensorFlow Hub (Google), PyTorch Hub (Meta): Model repositories maintained by framework vendors.
vLLM, TGI, NVIDIA NIM: Self-hosted LLM inference servers that compete with Hugging Face's Inference Endpoints, though TGI is part of the Hugging Face ecosystem.

Strategy

1. Own the open-source distribution layer

Hugging Face's strategic value lies in being the default platform where open models are shared, discovered, and deployed. Nvidia's acquisition underscores this: as major AI labs build their own chips, controlling the open-source ecosystem becomes a defensive moat. Nvidia gains visibility into customer preferences, trending models, downloaded datasets, and emerging architectures weeks before they hit mainstream tech news.

2. Monetize at the enterprise and infrastructure layer

Free access drives adoption; paid enterprise features, inference, and compute capture value from the small share of users who need production-grade capabilities. Inference Endpoints are the largest revenue line, followed by Enterprise Hub and Pro subscriptions.

3. Expand into robotics

The acquisition of Pollen Robotics and the launch of Reachy Mini and a $399 open-source robot duck represent an expansion beyond software into affordable open-source hardware, extending the open-source ethos to physical AI.

4. Maintain platform neutrality under Nvidia ownership

Nvidia has committed that Hugging Face "will remain an open platform for the entire AI ecosystem," with developers able to choose which models, cloud providers, and chips they use. This neutrality is central to the platform's value proposition and will be tested under Nvidia ownership.

The CODEW analysis

Independent analysis, separate from the factual record above. Draft for editorial review.

Nvidia's acquisition of Hugging Face is a defensive move as much as an offensive one. The largest closed-source AI labs — OpenAI, Google, Amazon, and Anthropic — are all building their own chips. By acquiring the central hub of open-source AI, Nvidia ensures its hardware remains the default target for the models that flow through Hugging Face. The deal also gives Nvidia data-driven foresight into what next-generation chip architectures they need to build based on developer habits. They are securing the software layer because hardware-only companies get commoditized over time.

The $12.93 billion price at 86x revenue is a strategic premium, not a financial one. Hugging Face's $150 million in annualized revenue does not justify a valuation approaching $13 billion on traditional metrics. Nvidia is paying for the platform's position at the center of the open-source AI ecosystem, where 18 million developers discover, share, and deploy models. The logic is closer to IBM's $34 billion acquisition of Red Hat in 2019 — a strategic buyer paying for a developer ecosystem position, not short-term cash flow.

The open-source neutrality question is now live. Hugging Face built its reputation as a neutral, community-driven platform. Under Nvidia ownership, competitors — including AI labs that build their own chips and cloud providers that compete with Nvidia — may seek alternatives. Nvidia has pledged to keep Hugging Face open and not require users to deploy on its silicon. How Nvidia manages this tension will determine whether Hugging Face retains its ecosystem gravity. Delangue has said the platform is "incentivised and motivated to keep this open" and noted that the deal "will allow AI to be more distributed all over the world and avoid too much concentration of power."

The deal will face regulatory scrutiny. Nvidia hopes to close by 2027, but the transaction is likely to face scrutiny from competition regulators given Nvidia's dominant position in AI chips. Nvidia's Justin Boitano has said, "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."

What to watch: whether Hugging Face remains model- and hardware-neutral under Nvidia; whether competing model hubs emerge, particularly in China; enterprise adoption trends for the paid tiers; the pace of open-source model releases post-acquisition; and regulatory developments in the U.S., EU, and U.K.

Related coverage

Company Intelligence
Enterprise AI Intelligence
Company Analysis and Company Deep Dive: Nvidia / AI Infrastructure
Watch Series: Open-Source AI
The Term Sheet: Nvidia's acquisition strategy
Related profiles: Nvidia, Alphabet / Google, Meta, Microsoft, Amazon / AWS, Anthropic, OpenAI

Sources

This profile is a durable reference page and is not investment advice. Factual sections are sourced above; the CODEW analysis section reflects The CODEW's independent view. Last updated October 15, 2026.

The CODEW