Chip comparison
NVIDIA Jetson AGX Thor 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 | NVIDIA Jetson AGX Thor | NVIDIA L40S |
|---|---|---|
| Process node | TSMC 4NP | TSMC 4N |
| Transistors (B) | — | 76.3 |
| TDP | 130 W | 350 W |
| Memory | 128 GB LPDDR5X | 48 GB GDDR6 |
| Memory bandwidth | 273 GB/s | 864 GB/s |
| FP16 (dense) | — | 362 TFLOPS |
| BF16 (dense) | — | 362 TFLOPS |
| FP8 (dense) | 2,070 TFLOPS | 733 TFLOPS |
| INT8 (dense) | 2,070 TOPS | 733 TOPS |
| Form factor | Module (AGX Thor) | PCIe |
| Announced | 2024-03-18 | 2023-08-08 |
| Released | 2025-08-01 | — |
Robots running NVIDIA L40S
No robots publicly running it yet.
Data centers with NVIDIA Jetson AGX Thor
No data centers publicly running it yet.
Data centers with NVIDIA L40S
No data centers publicly running it yet.
Common questions
NVIDIA Jetson AGX Thor vs NVIDIA L40S: which is faster for training?
NVIDIA Jetson AGX Thor has 2.82x the dense FP8 throughput of the other (NVIDIA Jetson AGX Thor: 2,070 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.
NVIDIA Jetson AGX Thor vs NVIDIA L40S: which has more memory?
NVIDIA Jetson AGX Thor carries more HBM (NVIDIA Jetson AGX Thor: 128 GB LPDDR5X; NVIDIA L40S: 48 GB GDDR6). Memory capacity sets the largest model that fits on one chip without partitioning; memory bandwidth sets inference throughput.
NVIDIA Jetson AGX Thor vs NVIDIA L40S: which is more power-efficient?
Dense FP8 TFLOPS per watt: NVIDIA Jetson AGX Thor 15.92 (2,070 ÷ 130 W); NVIDIA L40S 2.09 (733 ÷ 350 W). NVIDIA Jetson AGX Thor wins at the die level; system-level efficiency also depends on cooling and interconnect.
NVIDIA Jetson AGX Thor vs NVIDIA L40S: which is more widely used?
NVIDIA Jetson AGX Thor: 11 robots and 0 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 · NVIDIA Jetson AGX Thor full page · NVIDIA L40S full page.