GPU comparison
Inferentia2 vs Trainium1: specs and rental prices
AWS Inferentia2 (Inf2) against AWS Trainium (Trn1): 32 GB against 32 GB of memory, lowest listed on-demand price $0.76 against $1.34 per chip-hour. Specs from AWS, prices as of Oct 29, 2026.
In short
- Both list 32 GB of GPU memory.
- Memory type is HBM on the Inferentia2 and HBM on the Trainium1.
- The Inferentia2 lists 816.7 GB/s (derived) of memory bandwidth against 612.5 GB/s (derived) for the Trainium1: 1.33× higher.
- Not every figure can be compared: at least one vendor page does not publish: FP8 tensor throughput; TF32 tensor throughput; FP64 throughput; Maximum power.
- The lowest listed on-demand price is $0.76 per chip-hour for the Inferentia2 (AWS) and $1.34 per chip-hour for the Trainium1 (AWS).
- Per GB of GPU memory that is $0.024 per hour for the Inferentia2 and $0.042 for the Trainium1.
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 | Inferentia2 (Inf2) | Trainium1 Trainium (Trn1) | Difference |
|---|---|---|---|
| GPU memory | 32 GB | 32 GB | Same |
| Memory type | HBM | HBM | Same |
| Memory bandwidth | 816.7 GB/s (derived) | 612.5 GB/s (derived) | Inferentia2 1.33× higher |
| FP16 / BF16 tensor throughput | 190 TFLOPS (basis not stated) | 187.5 TFLOPS (basis not stated) (derived) | Not compared: the vendors state the figures on different bases |
| GPU-to-GPU interconnect | 192 GB/s | 768 GB/s | Trainium1 4× higher |
Compared versions: Inferentia2 (Inf2) and Trainium1 Trainium (Trn1). 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: Inferentia2 specifications · Trainium1 specifications.
Rental prices
| Rental price | Inferentia2 | Trainium1 | Difference |
|---|---|---|---|
| Lowest listed on-demand price | $0.76 per chip-hour · AWS · checked Oct 29, 2026 | $1.34 per chip-hour · AWS · checked Oct 29, 2026 | Trainium1 1.77× higher |
| Lowest listed spot price | No fresh listing | No fresh listing | — |
| Price per GB of memory | $0.024 per GB of GPU memory per hour | $0.042 per GB of GPU memory per hour | Trainium1 1.77× higher |
| Fresh listings | 2 from 1 providers | 2 from 1 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: Inferentia2 prices · Trainium1 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). Inferentia2 vs Trainium1. Retrieved from https://datacentercensus.com/gpu-prices/compare/inferentia2-vs-trainium1 |
|---|---|
| MLA | “Inferentia2 vs Trainium1.” Data Center Census, Oct 29, 2026, https://datacentercensus.com/gpu-prices/compare/inferentia2-vs-trainium1. |
| Link | https://datacentercensus.com/gpu-prices/compare/inferentia2-vs-trainium1 |
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.