Ranking · H100 · 80 GB
H100 cloud providers, ranked by price
19 providers with a public on-demand H100 price, ordered by their lowest listed price per GPU-hour. Data read 12 November 2026.
| # | Provider | $/GPU-hr | Billing | Plan | Source |
|---|---|---|---|---|---|
| 1 | Voltage Park | $1.99 | per hour | 1× | page |
| 2 | Vultr | $2.30 | per hour | 8× HGX | page |
| 3 | Hyperstack | $2.50 | per minute | 1× PCIe | page |
| 4 | Massed Compute | $2.73 | not stated | 1× | page |
| 5 | RunPodour pick | $2.89 | per second | 1× PCIe | page |
| 6 | Thunder Compute | $2.95 | per second | 1× SXM | page |
| 7 | Civo | $2.99 | per hour | 1× SXM | page |
| 8 | OVHcloud | $2.99 | per hour | 1× | page |
| 9 | Runcrateours, live | $3.58 | per second | 1× | page |
| 10 | Crusoe | $3.90 | per hour | 8× HGX | page |
| 11 | Together AI | $3.99 | not stated | 8× HGX | page |
| 12 | DigitalOcean (Paperspace) | $4.41 | per second | 1× HGX | page |
| 13 | Nebius | $4.50 | not stated | 8× HGX | page |
| 14 | Verda | $5.09 | not stated | 1× SXM | page |
| 15 | CoreWeave | $6.16 | not stated | 8× HGX | page |
| 16 | Amazon Web Services | $6.88 | per second | 1× | page |
| 17 | Microsoft Azure | $6.98 | not stated | 1× NVL | page |
| 18 | Oracle Cloud | $10.00 | not stated | 1× | page |
| 19 | Google Cloud | $11.06 | not stated | 8× | page |
By price
H100 price, provider by provider
- Voltage Park$1.99/GPU-hr
- Vultr$2.30/GPU-hr
- Hyperstack$2.50/GPU-hr
- Massed Compute$2.73/GPU-hr
- RunPod$2.89/GPU-hr
- Thunder Compute$2.95/GPU-hr
- Civo$2.99/GPU-hr
- OVHcloud$2.99/GPU-hr
- Runcrateours$3.58/GPU-hr
- Crusoe$3.90/GPU-hr
- Together AI$3.99/GPU-hr
- DigitalOcean (Paperspace)$4.41/GPU-hr
- Nebius$4.50/GPU-hr
- Verda$5.09/GPU-hr
- CoreWeave$6.16/GPU-hr
- Amazon Web Services$6.88/GPU-hr
- Microsoft Azure$6.98/GPU-hr
- Oracle Cloud$10.00/GPU-hr
- Google Cloud$11.06/GPU-hr
Questions
Who has the lowest H100 price?
Voltage Park at $1.99 per GPU-hour on the date shown, against a median of $3.90.
Is the lowest-priced provider the right one?
Not by itself. Price is one input. Availability, region, billing granularity, storage and egress rules and support differ, and none of them show in a price list.
How is this ranking made?
By one rule, stated above: lowest listed on-demand price per GPU-hour. There is no paid placement and no scoring beyond that.
Other GPUs
Disclosure
Provider Index is published by the Runcrate team, who also run Runcrate, an AI cloud. Runcrate is listed in the table for H100, marked "ours, live". It is sorted by price like every other row, with no pinned position. Other providers' prices are read from their own public pricing pages. We have no paid placement.