Key Partners
- AWS — Marketplace org billing; Inference Endpoints on AWS GPUs.
- Google Cloud — Vertex Model Garden, GKE, Cloud Run GPUs partnership.
- Microsoft Azure — named cloud partner path for HF in customer clouds.
- Open-model labs & community authors — fill Hub with weights and datasets.
- Inference Providers network — multi-provider serverless routing on Hub.
- GPU / accelerator vendors — underlying capacity for Spaces and Endpoints.
Key Activities
- Operate Hub storage, CDN, discovery, and collaboration (git ML repos).
- Maintain Spaces (Gradio/Docker/ZeroGPU) and Jobs for demos and training.
- Run Inference Providers (serverless) and Inference Endpoints (dedicated).
- Ship open-source libraries and Hub UX that deepen daily habits.
- Sell and onboard Team / Enterprise (SSO, SCIM, security, support).
- Partner with CSPs so Hub models deploy in customer clouds.
Key Resources
- Hub catalog — 3M+ public models; 500k+ datasets (Spring 2026 / Aug 2026).
- Community & brand — default place to share and discover open ML.
- Spaces directory — hundreds of thousands of AI apps/demos.
- Open-source libraries and Hub APIs that lock in developer workflow.
- Compute control plane — Providers, Endpoints, Jobs, ZeroGPU.
- Enterprise security stack — SSO/SCIM, access control, data location.
Value Propositions
- Git-based Hub for models, datasets, Spaces — the open ML catalog.
- Spaces — share interactive demos (CPU free; GPU PAYG).
- Inference Providers — serverless calls to 10k+ / 250k+ models (credits).
- Inference Endpoints — dedicated, autoscaling production deploy.
- Team/Enterprise — SSO, storage, billing controls, priority support.
- CSP paths — use Hub models inside AWS / Azure / Google Cloud.
Customer Relationships
- Community PLG — free Hub, Discussions, Posts, Articles, Spaces.
- Self-serve PRO — badge, credits, Dev Mode for serious individuals.
- Team — org seats, spending limits, Dataset Viewer on private data.
- Enterprise — custom onboarding, legal/compliance, dedicated support.
- Usage dashboards and credit auto-recharge for compute trust.
- AWS Marketplace / CSP billing for procurement-friendly orgs.
Channels
- Primary door — huggingface.co Hub (models, datasets, Spaces, orgs).
- Spaces as viral demos — shareable ML apps.
- Docs, blogs, and social Posts / Articles on profiles.
- In-product subscribe for PRO and Team.
- Enterprise sales + annual contracts.
- AWS Marketplace; Azure / Google Cloud partner surfaces.
Customer Segments
- Individual developers and researchers (Free + PRO).
- Startups and product teams collaborating on private models (Team).
- Enterprises needing SSO/SCIM, compliance, priority support.
- MLOps / platform teams deploying Inference Endpoints.
- Open-model labs publishing weights to reach users.
- CSP customers discovering Hub models inside cloud consoles.
Cost Structure
- Hub storage + egress/CDN for models and datasets (PB-scale).
- GPU/CPU capacity for Spaces, Jobs, Inference Endpoints, ZeroGPU.
- Inference Providers routing / partner compute economics.
- Engineering — Hub, libraries, security, billing systems.
- Enterprise sales, success, and dedicated support.
- Community programs, events, and open-source maintainership.
Revenue Streams
- PRO — $9/month individual subscription (FACT — /pro).
- Team — $20/user/month; Enterprise from $50/user/month (FACT — pricing).
- Spaces upgraded hardware — CPU Upgrade $0.03/hr; T4 from $0.40/hr.
- Inference Endpoints — dedicated deploy from ~$0.03/hr CPU; GPU SKUs higher.
- Inference Providers — serverless usage; credits included then PAYG.
- Private storage overage — ~$18/TB/mo base with volume discounts.
- CSP / AWS Marketplace — compute billed via cloud account.