Chip comparison
AWS Trainium 2 vs NVIDIA L40S
Side-by-side specs
Straight from each vendor's datasheet. Winner in bold: higher throughput, memory, or transistor count; lower power draw or price. Sparse (2:4) throughput roughly doubles the FP8/INT8 numbers shown.
| Field | AWS Trainium 2 | NVIDIA L40S |
|---|---|---|
| Process node | TSMC N5 | TSMC 4N |
| Transistors (B) | — | 76.3 |
| TDP | 500 W | 350 W |
| Memory | 96 GB HBM3 | 48 GB GDDR6 |
| Memory bandwidth | 2,900 GB/s | 864 GB/s |
| FP16 (dense) | — | 362 TFLOPS |
| BF16 (dense) | 667 TFLOPS | 362 TFLOPS |
| FP8 (dense) | 1,300 TFLOPS | 733 TFLOPS |
| INT8 (dense) | 1,300 TOPS | 733 TOPS |
| Form factor | AWS Neuron 2 (Trn2 UltraServer with 64 chips per unit) | PCIe |
| Announced | 2024-12-03 | 2023-08-08 |
| Released | 2025-01-01 | — |
Robots running AWS Trainium 2
No robots publicly running it yet.
Robots running NVIDIA L40S
No robots publicly running it yet.
Data centers with AWS Trainium 2
Data centers with NVIDIA L40S
No data centers publicly running it yet.
Common questions
AWS Trainium 2 vs NVIDIA L40S: which is faster for training?
AWS Trainium 2 has 1.77x the dense FP8 throughput of the other (AWS Trainium 2: 1,300 TFLOPS; NVIDIA L40S: 733 TFLOPS). Real-world training speed also depends on memory bandwidth, interconnect, and software stack: see the spec diff for those inputs.
AWS Trainium 2 vs NVIDIA L40S: which has more memory?
AWS Trainium 2 carries more HBM (AWS Trainium 2: 96 GB HBM3; NVIDIA L40S: 48 GB GDDR6). Memory capacity sets the largest model that fits on one chip without partitioning; memory bandwidth sets inference throughput.
AWS Trainium 2 vs NVIDIA L40S: which is more power-efficient?
Dense FP8 TFLOPS per watt: AWS Trainium 2 2.60 (1,300 ÷ 500 W); NVIDIA L40S 2.09 (733 ÷ 350 W). AWS Trainium 2 wins at the die level; system-level efficiency also depends on cooling and interconnect.
AWS Trainium 2 vs NVIDIA L40S: which is more widely used?
AWS Trainium 2: 0 robots and 3 data centers publicly running it. NVIDIA L40S: 0 robots and 0 data centers. Empty here means no operator has said publicly, not zero adoption in the market.
See also: every chip comparison · AWS Trainium 2 full page · NVIDIA L40S full page.