Hugging Face — research brief

# Hugging Face — research brief

**Slug:** hugging-face  
**Date:** 2026-09-20 (America/Chicago)  
**Industry:** AI / ML Platform (open model Hub + compute)

## Teaching arc
Open Hub (models/datasets/Spaces) → Inference Providers & Spaces hardware → Team / Enterprise Hub seats → Inference Endpoints + CSP partner paths (AWS / Azure / Google Cloud)

## FACT (primary)

| Claim | Source |
|-------|--------|
| PRO = $9/month; includes 1 TB private storage, 10 TB public, 20× inference credits, 8× ZeroGPU quota, Spaces Dev Mode | https://huggingface.co/pro |
| Team = $20 / user / month | https://huggingface.co/pricing |
| Enterprise = $50 / user / month (list) with SCIM, advanced security, managed billing, dedicated support | https://huggingface.co/pricing |
| Hub storage: public base $12/TB/mo → volume to $8/TB; private $18/TB → volume to $12/TB | https://huggingface.co/pricing |
| Spaces hardware: CPU Basic free; CPU Upgrade $0.03/hr; T4 small $0.40/hr; A100 large $2.50/hr; up to 8× A100 $20/hr | https://huggingface.co/pricing |
| Inference Endpoints: CPU from ~$0.03/hr (AWS Sapphire Rapids 1×); T4 $0.50/hr; H100 $4.50/hr; B200 $9.25/hr | https://huggingface.co/pricing |
| Monetization thesis: Hub collaboration platform + advanced features + simple access to compute | https://huggingface.co/docs/hub/en/billing |
| Team/Enterprise: SSO, access control, data location, higher storage; Team via card or AWS; Enterprise annual contract | https://huggingface.co/docs/hub/en/billing |
| Compute (Spaces, Inference Endpoints, Inference Providers) billed separately PAYG; credits from PRO/Team/Enterprise apply first | https://huggingface.co/docs/hub/en/billing |
| Cloud partners: AWS, Azure, Google Cloud; AWS Marketplace org linking for consolidated billing | https://huggingface.co/docs/hub/en/billing |
| Spring 2026: 13M users, 2M+ public models, 500k+ public datasets (2025 close) | https://huggingface.co/blog/huggingface/state-of-os-hf-spring-2026 |
| Hub crossed 3M public models (Aug 2026 HF staff blog / counter) | https://huggingface.co/blog/ivanfioravanti/three-million-models-and-counting |
| Google Cloud partnership: Vertex Model Garden, GKE, Cloud Run GPUs, Inference Endpoints on GCP | https://huggingface.co/blog/google-cloud |

## SECONDARY (press / CEO quotes — label as such)

| Claim | Source |
|-------|--------|
| Crossed ~$100M annual revenue run rate (CEO Clément Delangue, mid-2026) | Analytics India Magazine summarizing Delangue X post |
| Platform free/open for ~97% of users; stores/serves hundreds of PB | Same AIM piece quoting Delangue |
| Last disclosed valuation ~$4.5B (Aug 2023 round); ~$400M raised (Crunchbase via AIM) | AIM |
| Contrary Research: 2K+ paying enterprise customers (as of Jun 2025 report); freemium open-core | https://research.contrary.com/report/hugging-face |

## INFERENCE (not cited as fact)

- PLG land on free Hub → PRO for solo power users → Team for collab orgs → Enterprise for SSO/compliance → attach PAYG compute (Spaces / Endpoints / Providers).
- CSP partnerships both distribute (deploy HF models in customer cloud) and bill (Marketplace) — channels + partners + revenue.

## Do not invent
No unverified ARR split, headcount, or valuation beyond labeled SECONDARY.