On-demand, the H100 rents for $3.55/hr and the MI325X for $3.26/hr — the MI325X is ~9% cheaper per hour, averaged across cloud providers.
| H100 | MI325X | |
|---|---|---|
| On-demand /hr | $3.55 | $3.26 |
| Spot /hr | $1.59 | $— |
| Reserved /hr | $3.09 | $1.98 |
| Per month (on-demand) | $2,592 | $2,380 |
| H100 | MI325X | |
|---|---|---|
| Vendor | NVIDIA | AMD |
| Architecture | Hopper | CDNA 3 |
| Memory | 80 GB HBM3 | 256 GB HBM3e |
| Memory bandwidth | 3.35 TB/s | 6 TB/s |
| Launched | 2022 | 2025 |
On-demand, the H100 averages $3.55/hr vs the MI325X at $3.26/hr — the MI325X is about 9% cheaper per hour. Spot rates are $1.59/hr vs $—/hr.
H100 costs more per hour but offers 80 GB HBM3 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 MI325X is usually the better value.
At on-demand rates that's about $2,592/mo for the H100 and $2,380/mo for the MI325X (730 hours).
See full pricing & history: H100 price · MI325X price · all GPUs.
Cross-provider averages from public cloud GPU listings. Informational only.