Nvidia has announced the Jetson Orin Nano 2, a small single-board computer the company pitches as a "brain" for robots, drones and camera-based AI systems that need to think without phoning home to a data center.
The headline claim is performance: Nvidia says the board roughly doubles AI inference performance over its predecessor. According to 24/7 Wall St., it does so while drawing 40% less power — a combination that matters enormously for anything running on a battery, like a delivery or inspection drone.
Nvidia's own framing, reported by The Robot Report, is that "frontier intelligence has reached the edge" and can now run in real time. In plainer terms: the large, capable AI models that until recently needed racks of servers are increasingly able to run on hardware small enough to bolt onto a machine. SiliconANGLE describes the board as aimed at running frontier-level models at the edge, and Nvidia positions it as an entry-level product — the on-ramp for the millions of developers who build robots and vision systems rather than the specialists buying top-end silicon.
Partners are already lining up. Aptiv announced support for the Jetson Orin Nano 2 as part of its push toward what it calls production-ready physical AI.
One significant catch: German outlet heise reports that market launch is not until 2027, so this is an announcement of intent well ahead of anything shipping.
Why it matters: cheap, low-power hardware that can run modern AI models locally is the bottleneck standing between impressive robot demos and robots that actually work in warehouses, fields and skies — and Nvidia is moving to own that bottleneck at the entry level, where the largest number of developers start.