Key Partners
- NVIDIA — GPUs, Exemplar Cloud (GB200 NVL72), collaboration / investment news (FACT — IR + company news).
- Data-center and power providers for MW-scale AI factories (FACT — IR capacity).
- AI labs & enterprises: e.g. Cognition, Cursor, Midjourney, Runway, CrowdStrike (FACT — IR).
- Hyperscaler cloud customers (expanded relationships — FACT IR; Microsoft/OpenAI historically disclosed).
- Capital / asset-level financing secured by take-or-pay contracts (SECONDARY — investor materials).
- Platform M&A: OpenPipe, Marimo, Monolith (FACT — FY25 IR highlights).
Key Activities
- Deploy and operate NVIDIA GPU / CPU clusters at MW scale.
- Run Kubernetes-native cloud + Mission Control observability.
- Price and fulfill spot, on-demand, and reserved capacity.
- Negotiate multi-year take-or-pay enterprise / hyperscaler deals.
- Attach storage, networking, managed software, Federal.
- Expand active and contracted power (IR capacity milestones).
Key Resources
- NVIDIA GPU fleets (H100, H200, B200, GB200 NVL72, etc.).
- >850 MW active power · ~3.1 GW contracted (YE2025 FACT — IR).
- Mission Control + Kubernetes-native platform software.
- AI Object Storage, networking, managed services IP.
- Revenue backlog $66.8B (FACT — IR YE2025).
- Public company capital access (Nasdaq CRWV).
Value Propositions
- Purpose-built AI cloud on latest NVIDIA GPUs.
- Kubernetes-native DX + Mission Control reliability.
- Spot / on-demand land → reserved expand.
- Free egress + AI Object Storage for data gravity relief.
- Take-or-pay scale for labs and hyperscalers.
- Federal / managed software for regulated workloads.
Customer Relationships
- Self-serve Login + published pricing for on-demand/spot.
- Contact sales for reserved (up to 60% off) and custom.
- Solutions / field engineering for cluster design.
- 24/7 support claims for production AI workloads (homepage).
- Named pioneer accounts as reference relationships (IR).
- Expand path — hours → reserved → take-or-pay renewals.
Channels
- Primary — coreweave.com + /pricing rate card.
- Login / Contact us conversion paths.
- Direct enterprise & hyperscaler sales.
- NVIDIA / ISV partner co-sell and announcements.
- IR + newsroom for procurement credibility.
- Docs, status, Mission Control for operators.
Customer Segments
- AI labs / model companies (training + inference at scale).
- AI-native startups & platforms (Cursor, Midjourney, Runway — IR).
- Enterprises (e.g. CrowdStrike partnership — IR).
- Hyperscalers needing supplemental GPU capacity (IR).
- VFX / rendering and Physical AI workloads (site solutions).
- Government / CoreWeave Federal buyers.
Cost Structure
- GPU / server capex and PP&E depreciation (IR).
- Power, colocation, and DC leases.
- Technology & infrastructure opex ($2.929B FY2025 FACT).
- Cost of revenue / delivery ($1.453B FY2025 FACT).
- Interest on asset-level and parent financing.
- S&M + G&A (smaller vs infra; FACT IR lines).
Revenue Streams
- Spot GPU hours — discounted, interruptible (FACT rates).
- On-demand PAYG — e.g. H100 8× $49.24/hr; B200 $68.80 (FACT).
- Inference single-GPU lane — e.g. H100 $6.16/hr (FACT).
- Reserved / committed — up to 60% off on-demand (FACT).
- Multi-year take-or-pay — backlog $66.8B YE2025 (FACT — IR).
- Storage (Hot $0.06/GB-mo…) + networking + managed attach (FACT).
- FY2025 revenue $5.131B (FACT — IR).