On-demand, the H200 rents for $4.37/hr and the MI355X for $8.60/hr — the H200 is ~97% cheaper per hour, averaged across cloud providers.
| H200 | MI355X | |
|---|---|---|
| On-demand /hr | $4.37 | $8.60 |
| Spot /hr | $2.98 | $5.29 |
| Reserved /hr | $3.72 | $— |
| Per month (on-demand) | $3,190 | $6,278 |
| H200 | MI355X | |
|---|---|---|
| Vendor | NVIDIA | AMD |
| Architecture | Hopper | CDNA 4 |
| Memory | 141 GB HBM3e | 288 GB HBM3e |
| Memory bandwidth | 4.8 TB/s | 8 TB/s |
| Launched | 2024 | 2025 |
On-demand, the H200 averages $4.37/hr vs the MI355X at $8.60/hr — the H200 is about 97% cheaper per hour. Spot rates are $2.98/hr vs $5.29/hr.
MI355X costs more per hour but offers 141 GB HBM3e memory and newer architecture — for large-model training the higher hourly rate often yields a lower cost-per-job. For memory-bound inference or budget runs, the H200 is usually the better value.
At on-demand rates that's about $3,190/mo for the H200 and $6,278/mo for the MI355X (730 hours).
See full pricing & history: H200 price · MI355X price · all GPUs.
Cross-provider averages from public cloud GPU listings. Informational only.