Battery drone with a 15 W compute ceiling: which Jetson production modules have a max power envelope of 15 W or less?

Question

Rapidflare · Accepted Answer

Six Jetson production modules have a default-mode power envelope of 15 W or less, so they can all operate within your 15 W compute ceiling:

| Module | Power envelope (default modes) | GPU arch | AI perf. (default) | Memory |
|--------|------|------|------|------|
| Jetson Orin Nano 8GB | 7–15 W | Ampere, 1024 CUDA + 32 Tensor cores | 40 INT8 TOPS | 8 GB LPDDR5, 68 GB/s |
| Jetson Orin Nano 4GB | 7–10 W | Ampere, 512 CUDA + 16 Tensor cores | 20 INT8 TOPS | 4 GB LPDDR5, 34 GB/s |
| Jetson TX2 | 7.5–15 W | Pascal, 256 CUDA cores | 0.666 TFLOPS FP32 | 8 GB LPDDR4, 59.7 GB/s |
| Jetson TX2 4GB | 7.5–15 W | Pascal, 256 CUDA cores | 0.666 TFLOPS FP32 | 4 GB LPDDR4, 51.2 GB/s |
| Jetson TX2 NX | 7.5–15 W | Pascal, 256 CUDA cores | 0.666 TFLOPS FP32 | 4 GB LPDDR4, 51.2 GB/s |
| Jetson Nano | 5–10 W | Maxwell, 128 CUDA cores | 0.236 TFLOPS FP32 | 4 GB LPDDR4, 25.6 GB/s |

A few points relevant to a battery drone:

- **Best performance within the ceiling:** The Jetson Orin Nano 8GB is the clear front-runner — 40 INT8 TOPS at 7–15 W in its default power modes, roughly an order of magnitude more AI throughput than the Pascal- and Maxwell-based options.
- **Super mode caveat:** The Orin Nano modules support a Super mode that raises their envelope to 7–25 W — above your ceiling — so on this drone you'd run them in their default modes, where the 8GB does 40 TOPS and the 4GB does 20 TOPS.
- **Lifecycle note:** The Jetson TX2 and TX2 4GB are marked end-of-life in the catalog, so for a new design the Orin Nano 8GB, Orin Nano 4GB, and Jetson Nano are the candidates to focus on.
- All six are 260-pin SO-DIMM (69.6×45 mm) or 400-pin B2B form factors, so they mount on compact carrier boards suited to drone payloads.

If you'd like, I can also check partner carrier boards or complete onboard computers built on these modules for your airframe.

Jetson production modules with a max power envelope of 15 W or less

Sources

Module	Power envelope (default modes)	GPU arch	AI perf. (default)	Memory
Jetson Orin Nano 8GB	7–15 W	Ampere, 1024 CUDA + 32 Tensor cores	40 INT8 TOPS	8 GB LPDDR5, 68 GB/s
Jetson Orin Nano 4GB	7–10 W	Ampere, 512 CUDA + 16 Tensor cores	20 INT8 TOPS	4 GB LPDDR5, 34 GB/s
Jetson TX2	7.5–15 W	Pascal, 256 CUDA cores	0.666 TFLOPS FP32	8 GB LPDDR4, 59.7 GB/s
Jetson TX2 4GB	7.5–15 W	Pascal, 256 CUDA cores	0.666 TFLOPS FP32	4 GB LPDDR4, 51.2 GB/s
Jetson TX2 NX	7.5–15 W	Pascal, 256 CUDA cores	0.666 TFLOPS FP32	4 GB LPDDR4, 51.2 GB/s
Jetson Nano	5–10 W	Maxwell, 128 CUDA cores	0.236 TFLOPS FP32	4 GB LPDDR4, 25.6 GB/s