Directory
Compare GPU Cloud Providers
Compare GPU cloud providers offering H100, H200, B200 and other accelerators: regions, pricing notes, availability and where to sign up.
Looking for rates? See the GPU Cloud Price Index or the full GPU price tracker.
| Provider | GPU models | Regions | Pricing | Availability | Website |
|---|---|---|---|---|---|
| Amazon Web Services | H100, H200, B200 | us-east-1, us-west-2, eu-west-1, ap-northeast-1 | On-demand, savings plans, capacity blocks | Capacity blocks and reservations | Visit → |
| CoreWeave | H100, H200, B200, GB200 | US East, US West, Europe | Reserved and on-demand contracts | Large dedicated clusters | Visit → |
| CUDO Compute | H100 PCIe, H100 SXM, H200, B200, B300, GB200, GB300 | — | Enterprise quote-based dedicated GPU infrastructure | Dedicated GPU clusters and enterprise AI infrastructure | Visit → |
| DigitalOcean | H100, H200, B300, MI300X, MI325X, MI350X, L40S | New York, Amsterdam, Toronto, Atlanta | On-demand, spot and 12-month reserved GPU plans | Regional on-demand GPU Droplets and contract capacity | Visit → |
| Google Cloud | H100, H200, B200 | us-central1, europe-west4, asia-southeast1 | On-demand, committed use, dynamic scheduling | Reservation based | Visit → |
| Hyperstack | A100, H100, H200, B200, B300, L40 | — | On-demand, spot and dedicated cluster pricing | On-demand VMs and dedicated GPU clusters | Visit → |
| Lambda | H100, H200, B200 | US West, US East | Hourly on-demand and reserved | On-demand and reserved | Visit → |
| Microsoft Azure | H100, H200, GB200 | East US, West Europe, Sweden Central | On-demand and reserved | Quota based | Visit → |
| Nebius | H100, H200, B200 | Finland, Europe, US | On-demand and reserved | European capacity | Visit → |
| Oracle Cloud Infrastructure | H100, H200, B200 | US Midwest, Germany, Japan | Bare-metal GPU clusters | Cluster reservations | Visit → |
| OVHcloud | H100, H200, A100, L40S, L4 | Europe, North America, Asia-Pacific | Public cloud hourly pricing and monthly estimates | Public Cloud GPU instances | Visit → |
| Runpod | H100, H200, B200, B300, A100, MI300X | 30+ global regions | Per-second Pods, Serverless, clusters and reserved clusters | On-demand, community cloud and reserved capacity | Visit → |
| Scaleway | H100 | Paris, Amsterdam, Warsaw | Hourly on-demand | Regional European capacity | Visit → |
| Vast.ai | H100, H200, B200, B300, A100 | Global marketplace | Marketplace pricing varies with live supply and demand | Marketplace capacity varies by host and region | Visit → |
| Verda | H100, H200, B200, B300, GB300, A100, RTX PRO 6000 | Europe, United States, Asia | Pay-as-you-go, spot and reserved GPU pricing | Self-service GPU instances and multi-node clusters | Visit → |
| Vultr | H100, A100, L40S | 33+ global cloud data center regions | Pay-as-you-go cloud GPU and cluster pricing | Self-service GPU cloud and H100 clusters | Visit → |
Choosing a GPU cloud provider
GPU cloud providers range from hyperscalers such as AWS, Microsoft Azure and Google Cloud to specialist "neoclouds" built around NVIDIA accelerators. Hyperscalers offer broad managed services and global regions; specialists often provide faster access to the newest GPUs, larger contiguous clusters and simpler pricing.
When comparing providers, look at available GPU models (H100, H200, B200, GB200), cluster networking, storage throughput, contract terms, data-residency options and support. Our price tracker lists published hourly rates.