NVIDIA has taken the wraps off the Jetson Orin Nano 2, a new entry-level robotics computer that doubles the AI inference performance of last year’s Orin Nano Super in the same footprint while cutting power by 40 percent, and named Wing, Matic Robotics, Doosan Bobcat and Cognex as first movers.
78 TOPS, 8GB and Nemotron on-device
The Orin Nano 2 packs 78 trillion operations per second of AI compute, 8GB of memory and an 8-core Arm CPU into the same compact SoM form factor as its predecessor. NVIDIA credits upgraded Tensor Cores and higher memory bandwidth for the 2× inference gain, with a 15-watt mode delivering the previous generation’s performance at 40% less power. The module runs the full NVIDIA robotics software stack, including Jetson agent skills, and is optimised to run on-device large- and vision-language models such as NVIDIA Cosmos, Nemotron, Gemma 4 and Qwen 3. “Today’s small and medium frontier models have reached the accuracy of last year’s largest frontier models,” said Deepu Talla, NVIDIA’s VP of robotics and edge AI. Modules and dev kits arrive in the first half of 2027.
Wing, Matic, Cognex and Doosan Bobcat lead adoption
Wing, the Alphabet drone-delivery subsidiary already running Jetson Orin Nano Super in its fleet, plans to evaluate the Orin Nano 2 to improve real-time perception and reduce power draw. “Drone delivery depends on AI that can enable fast, reliable understanding of the real world,” said Wing head of perception Dinuka Abeywardena. Consumer robot maker Matic will use the new module for conversational AI, gesture detection and semantic understanding on its home cleaning robots, according to co-founder and CEO Navneet Dalal. Doosan Bobcat and machine-vision leader Cognex round out the first named enterprise adopters, alongside a partner network including Aptiv, ADLINK, Advantech, Antmicro, AAEON, Aetina, Connect Tech, RidgeRun, Seeed Studio and YUAN building carrier boards, systems and reference designs.
Filling the ‘entry-level’ Jetson tier as physical AI scales
NVIDIA now counts more than 3 million developers on its robotics stack. Orin Nano 2 fills out the low end of a Jetson roadmap that Jensen Huang has been marketing as the on-device layer of the company’s physical-AI push. The launch dovetails with this week’s AWS partnership expansion for 2 million more GPUs, the Cloverleaf AI-factory tie-up and Generalist AI’s $200M NVIDIA-backed round.
Reporting based on coverage from NVIDIA, GlobeNewswire and Robotics 24/7.