CoreWeave
thecontinuitylayer.com/tools/gpu-hosting/coreweaveCompanies running open models at large scale that need whole GPU servers and clusters
GPU cloud · CoreWeave
CoreWeave is a large specialist GPU cloud that rents full NVIDIA servers and clusters, mainly to AI labs and enterprises. Its pricing is listed per instance, where one instance is usually an 8-GPU server.
It suits teams that already run Kubernetes and need many GPUs for serving or training. It is not built for someone who wants a single GPU for an afternoon, and the best rates come with long contracts arranged through sales.
Specs
| Price per hour | HGX H100 (8 GPUs) $49.24/hr; HGX B200 (8 GPUs) $68.80/hr on demand |
|---|---|
| GPUs | Large NVIDIA clusters: HGX H100, H200, B200 and GB200 NVL72 systems |
| HGX H100 | $49.24/hr per 8-GPU instance on demand |
| HGX B200 | $68.80/hr per 8-GPU instance on demand |
| GB200 NVL72 | $42.00/hr per 4-GPU instance |
| Spot | Offered on some instance types |
| Platform | Kubernetes-based |
Good for
- Large-scale inference fleets
- Multi-node training
- Enterprises with committed spend
Not for
- Single-GPU experiments
- Individuals and small hobby projects
See also
- Together AI · Teams that want a dedicated, managed endpoint for an open model without running servers
- Novita AI · Developers who want both an open-model API and rentable GPUs from one account
- GPU Droplets · Developers already on DigitalOcean who want a GPU server next to their app
We last reviewed CoreWeave on 31 October 2026. Prices and specs change; check coreweave.com before you buy or sign up. Figures we could not confirm on the vendor's own page read “not published” or “varies”.
Vendor source: https://www.coreweave.com/pricing