H200 SXM specs
NVIDIA · Shipping now · first available Jun 2024 (estimated). Dense compute, memory, bandwidth and power from the vendor's documents; each precision stays separate.
All accelerators · Compare with H100 NVL
Specifications
| Dense BF16 · FP32 accumulate | 989 TFLOPS |
|---|---|
| Dense FP8 · FP32 accumulate | 1,979 TFLOPS |
| Dense FP4 · FP32 accumulate | Not published |
| Memory | 141 GB · HBM |
| Memory bandwidth | 4.8 TB/s |
| Maximum chip power | 700 W |
| Interconnect | NVLink 4 · 900 GB/s |
| Process node | TSMC 4N |
| First availability | Jun 2024est — estimated / secondary-source figure |
| Availability | Shipping |
Same compute as H100; HBM3e lifts capacity 1.8× and bandwidth 1.4×.
Dense tensor rates with FP32 accumulation, never sparse or FP16-accumulate headline figures. Memory is in decimal GB. The est marker flags a value its vendor-document review did not confirm.
Rent it
$3.99/GPU-hrCheapest on-demand · Hyperstack · observed
8 providers: AWS, Azure, Hyperstack, Jarvislabs, Nebius, Oracle OCI, Together AI, Verda. Quotes checked through .
H200 rental prices from every provider →Value ratios
| $ per dense BF16 PFLOP-hour | $4.03 |
|---|---|
| Dense BF16 TFLOPS per watt | 1.41 |
| Bandwidth bytes per 1,000 BF16 FLOP | 4.85 |
| $ per dense FP8 PFLOP-hour | $2.02 |
| Dense FP8 TFLOPS per watt | 2.83 |
| Bandwidth bytes per 1,000 FP8 FLOP | 2.43 |
| $ per GB of memory per hour | $0.028 |
Prices divide the cheapest on-demand list price; theoretical ratios do not measure application speed.
Sources
Specifications reviewed . SXM column; sparse tensor figures divided by two, with existing whole-TFLOPS rounding retained.
Rental price: Hyperstack published pricing ↗ (NVIDIA H200 SXM 141GB; maximum per GPU: 22 pCPUs and 225 GB RAM; actual allocation not specified; published per GPU-hour).