NVIDIA
NVIDIA H100
The Hopper-generation accelerator that became the standard unit of large-scale AI training capacity.
- Architecture
- Hopper
- Memory
- 80 GB HBM3 (SXM) / 80 GB HBM2e (PCIe)
- Memory bandwidth
- Up to 3.35 TB/s
- Release
- Announced 2022
Main use cases
Large language model training, inference, HPC
Cloud pricing
Full tracker →| Provider | Price / hour | Region | Availability | Updated |
|---|---|---|---|---|
| AWS | Not verified | us-east-1 | Capacity blocks | 28 Sept 2026, 17:19 |
| CoreWeave | Not verified | US East | Contract / reserved | 28 Sept 2026, 17:19 |
| DigitalOcean | USD 4.41 | — | On-demand GPU Droplet; per-GPU hourly rate | 29 Sept 2026, 07:24 |
| Lambda | USD 4.29 | — | 1-GPU on-demand list rate; tax extra; capacity not checked | 7 Oct 2026, 06:10 |
| Lambda | USD 3.29 | — | 1-GPU on-demand list rate; tax extra; capacity not checked | 7 Oct 2026, 06:10 |
| Nebius | USD 4.5 | — | Published on-demand USD rate per GPU-hour, effective 1 October 2026; taxes extra; capacity not checked | 7 Oct 2026, 06:04 |
| Runpod | USD 3.49 | — | Secure Cloud Pod; published on-demand rate | 29 Sept 2026, 07:24 |
| Runpod | USD 2.89 | — | Secure Cloud Pod; published on-demand rate | 29 Sept 2026, 07:24 |
| Scaleway | EUR 2.73 | — | Published per-GPU hourly price | 29 Sept 2026, 07:43 |
| Scaleway | EUR 6.61 | — | Published per-GPU hourly price for H100-SXM configuration | 29 Sept 2026, 07:43 |
| Verda | USD 3.25 | — | On-demand GPU instance; published rate | 29 Sept 2026, 07:28 |
Widely rented by the hour from GPU clouds; on-demand rates vary heavily by provider and commitment.
NVIDIA H100 in the news
GPU news →No coverage yet.