NVIDIA Jetson Orin Nano 2 Doubles AI Compute to 78 TOPS at 40W

NVIDIA has released the Jetson Orin Nano 2, a new edge AI module that doubles computational power while offering significantly better efficiency at lower power settings. This update matters because it gives robotics developers and edge AI builders a more capable platform for running complex models without increasing their energy footprint. The hardware targets entry-level applications where compact size and thermal constraints are critical.

Edge AI module doubles performance while cutting power consumption

The module keeps the Ampere architecture and an eight-core Arm CPU found in the previous generation. It pairs these components with 8 GB of LPDDR5x memory and delivers 120 GB/s of bandwidth. NVIDIA positions this device for entry-level edge AI and robotics tasks, maintaining a small form factor suitable for embedded systems.

Specifications

• Architecture: Ampere
• CPU: 8-core Arm CPU
• Memory: 8 GB LPDDR5x
• AI Compute: 78 TOPS
• TDP: 40W

At its maximum 40W thermal design power, the Jetson Orin Nano 2 delivers up to 78 TOPS of AI compute. This figure represents a 2x increase in performance compared to its predecessor. The company also highlights that running the device at 15W allows it to match the previous generation's performance while consuming 40% less power.

The platform supports several key models including NVIDIA Cosmos, Nemotron, Gemma 4, and Qwen 3. NVIDIA plans to make both the Developer Kit and the module available globally in the first half of 2027. Buyers looking to upgrade their edge AI infrastructure can expect a more efficient solution when shipments begin.

We tracked the earlier Jetson Orin Nano to understand the baseline for this generation's improvements. The new module offers a clear path for developers who need more AI throughput without redesigning their thermal systems. This release confirms NVIDIA's strategy to enhance efficiency alongside raw performance in the edge segment.