How does CoreWeave print?
GPU hours. Reserved clusters. The customer is a lab or a cloud that ran out.
A neo-cloud. Power and the rack are the product.
The mix
- GPU cloudHours and reserved.88%
- OtherCPU, storage, services.12%
Who pays
AI labs, funds, and anyone who cannot wait on a hyperscaler.
Who builds
Infra, facilities, and the people who keep the GPUs billed.