The H200 is the memory-bound sibling of the H100 — same compute, but 141GB of HBM3e at 4.8 TB/s. This makes it the strongest option for long-context inference, MoE models, and anything that stalls on memory pressure.
Specs
| Spec | Value |
|---|---|
| Architecture | Hopper |
| VRAM | 141 GB HBM3e |
| Memory bandwidth | 4.8 TB/s |
| FP16 (with sparsity) | 1,979 TFLOPS |
| TDP | 700W |
| Interconnect | 4th-gen NVLink 900 GB/s |
// LIVE_LISTINGS
NVIDIA H200 hosting from 0 providers
No live NVIDIA H200 listings right now — but we can source them.
Request a quoteNeed help sizing your NVIDIA H200 cluster?
Tell us your workload and we'll route matched offers from providers with real inventory — usually within 24 hours.
Get matched quotes