- GPU model
- NVIDIA H200
- Deployment
- Bare metal
- Region
- Europe
- Interconnect
- InfiniBand NDR 400G
Price on request
Service class
GPU Cloud
Virtualised, bare-metal or managed Kubernetes capacity. You rent compute; diligence is fabric, storage tier and exit terms.
What you are buying
You rent capacity for a term, or contract future availability. No hardware changes hands.
Check before you commit
- Virtualised, bare metal or managed Kubernetes — they are not interchangeable
- Interconnect fabric, which decides whether multi-node training scales
- Storage tier and throughput against your checkpoint pattern
- Egress charges and exit terms before you commit, not after
Need a specific configuration?
Tell us the model, quantity, region and timeline. A structured request reaches the sourcing team with full technical context, which gets a faster and more accurate answer than an open message.
Request capacityReserve future capacity How the process worksRented GPU capacity, sized to your workload.
Available now
Listings in this class
Verified inventory from approved suppliers. Every entry states its commercial model, and contracts are agreed directly with the supplier.
- GPU model
- AMD MI300X
- Region
- North America
Price on request