GPU comparison
Trainium2 vs H100: specs and rental prices
AWS Trainium2 (Trn2) against NVIDIA H100: 96 GB against 80 GB of memory. Specs from AWS and NVIDIA, prices as of Oct 29, 2026.
In short
- The Trainium2 lists 96 GB of GPU memory against 80 GB for the H100: 1.2× higher.
- Memory type is HBM3 on the Trainium2 and HBM3 on the H100.
- The H100 lists 3,350 GB/s of memory bandwidth against 2,875 GB/s (derived) for the Trainium2: 1.17× higher.
- Not every figure can be compared: at least one vendor page does not publish: FP16 / BF16 tensor throughput; TF32 tensor throughput; FP64 throughput; Maximum power; GPU-to-GPU interconnect.
- Only the H100 has a fresh on-demand listing ($2.50 per GPU-hour at Hyperstack); the Trainium2 has none right now.
This summary is generated from the tables below. Prices were last checked Oct 29, 2026. A listing not re-checked for 7 days is greyed and left out of every lowest-price figure; after 14 days it is removed from the page.
Specifications side by side
| Specification | Trainium2 (Trn2) | H100 SXM | Difference |
|---|---|---|---|
| GPU memory | 96 GB | 80 GB | Trainium2 1.2× higher |
| Memory type | HBM3 | HBM3 | Same |
| Memory bandwidth | 2,875 GB/s (derived) | 3,350 GB/s | H100 1.17× higher |
| FP8 tensor throughput | 1,300 TFLOPS (basis not stated) (derived) | 3,958 TFLOPS (with sparsity) | Not compared: the vendors state the figures on different bases |
| FP16 / BF16 tensor throughput | Not published | 1,979 TFLOPS (with sparsity) | — |
| TF32 tensor throughput | Not published | 989 TFLOPS (with sparsity) | — |
| FP64 throughput | Not published | 34 TFLOPS | — |
| Maximum power | Not published | 700 W | — |
| GPU-to-GPU interconnect | Not published | 900 GB/s | — |
Compared versions: Trainium2 (Trn2) and H100 SXM. The figure in the stronger cell is shaded; power is shown without a winner. Two throughput figures are compared only when both vendors state them on the same basis (both with sparsity, or both dense), and “derived” means a total the vendor publishes divided by the GPU count it publishes. Full tables with sources: Trainium2 specifications · H100 specifications.
Rental prices
| Rental price | Trainium2 | H100 | Difference |
|---|---|---|---|
| Lowest listed on-demand price | No fresh listing | $2.50 per GPU-hour · Hyperstack · checked Oct 29, 2026 | — |
| Lowest listed spot price | No fresh listing | $0.79 per GPU-hour · Nebius · checked Oct 29, 2026 | — |
| Price per GB of memory | Not available | $0.031 per GB of GPU memory per hour | — |
| Fresh listings | 0 from 0 providers | 26 from 12 providers | — |
Prices cover every version and instance size the providers list, so they are not tied to the versions compared above; the price per GB of memory divides the lowest fresh on-demand price by the memory of the compared version. Listed prices are not like for like across providers. See every listing: Trainium2 prices · H100 prices.
More on these GPUs
Cite this page
Last updated Oct 29, 2026. Figures may change as records are updated — include the date you retrieved them.
| APA | Data Center Census. (2026). Trainium2 vs H100. Retrieved from https://datacentercensus.com/gpu-prices/compare/trainium2-vs-h100 |
|---|---|
| MLA | “Trainium2 vs H100.” Data Center Census, Oct 29, 2026, https://datacentercensus.com/gpu-prices/compare/trainium2-vs-h100. |
| Link | https://datacentercensus.com/gpu-prices/compare/trainium2-vs-h100 |
Republishing a figure or table? Attribute it to Data Center Census with a link to this page. The underlying data is downloadable as JSON and CSV; the press kit has standing figures and contact details.