The NVIDIA H100 (Hopper, 80 GB HBM3) rents from $2.40/hr, averaged across 63 cloud providers, updated 2026-11-15. On-demand pricing has risen 23% over the past 90 days.
The workhorse of the current AI build-out — strong FP16/FP8 training and inference, and the most widely available data-center GPU. Hopper · 80 GB HBM3 · 3.35 TB/s memory bandwidth · launched 2022.
▮ on-demand /hr · ▮ spot /hr
| Date | On-demand | Spot |
|---|---|---|
| 2026-11-15 | $4.01 | $2.40 |
| 2026-11-14 | $4.00 | $2.06 |
| 2026-11-13 | $4.00 | $2.46 |
| 2026-11-12 | $3.99 | $2.46 |
| 2026-11-11 | $3.50 | $2.21 |
| 2026-11-10 | $3.99 | $2.46 |
| 2026-11-09 | $3.99 | $2.29 |
| 2026-11-08 | $3.99 | $2.47 |
| 2026-11-07 | $4.00 | $2.33 |
| 2026-11-06 | $3.99 | $2.46 |
| 2026-11-05 | $3.88 | $2.37 |
| 2026-11-04 | $3.74 | $2.13 |
| 2026-11-03 | $3.75 | $2.46 |
| 2026-11-02 | $3.79 | $2.42 |
| GPU | On-demand /hr | Spot /hr | Memory | |
|---|---|---|---|---|
| H100 | $4.01 | $2.40 | 80 GB HBM3 | |
| B200 | $8.28 | $4.80 | 192 GB HBM3e | H100 vs B200 → |
| H200 | $4.59 | $3.36 | 141 GB HBM3e | H100 vs H200 → |
| MI355X | $8.25 | $3.56 | 288 GB HBM3e | H100 vs MI355X → |
| MI325X | $3.11 | $— | 256 GB HBM3e | H100 vs MI325X → |
| MI300X | $2.86 | $1.84 | 192 GB HBM3 | H100 vs MI300X → |
The H100 averages $4.01/hr on-demand, $2.40/hr spot, and $3.26/hr reserved across 63 cloud providers (updated 2026-11-15).
The lowest tier is spot at about $2.40/hr on average; reserved (1–3 year commitment) runs ~$3.26/hr. Compare live per-provider rates on the Bargo GPU price tracker.
At on-demand $4.01/hr, one H100 runs about $2,927/month (730 hours); reserved pricing lowers that to roughly $2,380/month.
On-demand, the H100 is $4.01/hr vs the B200 at $8.28/hr — cheaper per hour. See the full H100 vs B200 comparison for specs and cost-per-job context.
Spot ($2.40/hr) is interruptible spare capacity at a discount to on-demand ($4.01/hr); use spot for checkpointed or fault-tolerant jobs and on-demand or reserved for long, uninterrupted runs.
Cross-provider averages from public cloud GPU listings. Informational only.