NVIDIA GPUs, rented or purpose-built.
GPU Cloud for hourly on-demand capacity, and Cluster Build for dedicated clusters we design, deploy, and operate. Both run on B300, B200, GB300, and H200 SXM.
Two ways to get GPUs running.
Need GPUs now? Rent them by the hour on GPU Cloud. Need dedicated infrastructure, a specific topology, or data residency guarantees? We'll build it for you. Both run the same GPU line-up.
Rent capacity
Pick a GPU in the portal and start. On-demand bills hourly; reserved instances lower the rate for longer runs.
Build dedicated
One team handles GPU selection, fabric design, power and cooling, installation, and operations for a dedicated cluster.
Blackwell and Hopper — pick the generation you need.
NVIDIA B300
Largest memory per node — fewer nodes for big models.
NVIDIA B200
The default Blackwell choice, used for both training and inference.
NVIDIA GB300
Delivered as NVL rack units, allocated by reservation only.
NVIDIA H200 SXM
Stable supply and lower rates — a good fit for inference.
GPUs alone don't make a training run.
A narrow fabric leaves GPUs idle; slow storage stalls checkpoints. Both products include the layers below.
Node-to-node traffic runs over InfiniBand. Topology depends on scale and communication pattern.
Checkpoints and training data on NVMe-backed PFS; archives on S3-compatible Object Storage.
Data ingress and egress ride the PacketStream backbone and direct ISP interconnects.
GPU, node, and fabric metrics exported to Prometheus, Grafana, Datadog, or your existing stack.
Handed over at the OS layer, or with Kubernetes and a scheduler configured.
Use Cases
Model training
Inference serving
Data processing
Regulated environments
Not sure which one fits?
Tell us the workload, how many GPUs you need, and for how long — we'll work out which option costs less.