NVIDIA Jetson Orin Nano Super Developer Kit

NVIDIA Jetson Orin Nano Super Developer Kit

The Jetson Orin Nano Super Developer Kit is an entry-level edge AI development platform in NVIDIA's Jetson Orin family, priced at $249. It targets developers, researchers, and students who need a capable, low-power AI compute system that runs inference workloads locally rather than relying on cloud infrastructure.

Hardware Architecture

The module is built around a 1024-core Ampere GPU with 32 Tensor Cores, which provides up to 67 TOPS of INT8 performance — the primary metric for neural network inference throughput. It is paired with 8 GB of LPDDR5 unified memory, shared between the CPU and GPU. The entire system operates within a configurable 7–25 W power envelope, making it suitable for embedded and battery-aware deployments.

Notably, existing Jetson Orin Nano hardware can reach the "Super" performance level through a software update, meaning the performance improvement does not require new silicon.

What It Can Run

The 67 TOPS figure is sufficient to execute a range of modern AI workloads at the edge:

• Small large language models (LLMs) and vision-language models, run entirely on-device
• Vision transformers (ViTs) for image classification, object detection, and segmentation
• Robotics perception pipelines, combining sensor fusion, depth estimation, and real-time control

Software Stack

The kit runs JetPack, NVIDIA's Linux-based SDK for Jetson, which bundles CUDA, TensorRT, cuDNN, and other acceleration libraries. Because this software stack is shared across the entire Jetson Orin lineup — including the more powerful Orin NX and AGX Orin modules — models and pipelines developed on the Nano Super can be migrated to higher-tier hardware without rewriting application code.

Typical Use Cases and Users

• Students and educators learning embedded AI, computer vision, and robotics
• Makers and hobbyists building smart camera systems, autonomous vehicles, or voice AI devices
• Engineers and researchers prototyping edge AI applications before committing to a production module
• Small-scale generative AI pilots where on-device inference is preferred for latency, privacy, or connectivity reasons