Nvidia has rolled out the Jetson Orin Nano 2—a compact AI computer built for robotics and edge AI. According to Nvidia, this little powerhouse delivers twice the inference performance of its predecessor, all while keeping the same small footprint. It can even run the latest language and multimodal models right on the device—no cloud needed.

What’s New with Jetson Orin Nano 2

The Jetson Orin Nano 2 sticks to the ultra-compact form factor the series is known for, but now brings a major speed boost: Nvidia claims a 2x jump in inference performance without changing the size. That means you can drop it into existing projects where local data crunching is a must.

Performance and Architecture

Under the hood, you get an 8-core Arm CPU and 8GB of RAM. AI compute tops out at 78 TOPS. Nvidia credits the speedup to improved Tensor Cores and beefed-up memory bandwidth, letting the module handle heavier AI workloads at the edge.

On-Device AI Capabilities

The Jetson Orin Nano 2 is built to run modern language and multimodal AI models locally, making it perfect for robotics and edge AI setups that need to work autonomously—without being tethered to the cloud 24/7.