Nvidia has introduced the Jetson Orin Nano 2, a compact robotics computer designed to run AI workloads at the edge. The company is targeting robots, delivery and inspection drones, and vision AI systems.
Smaller and more efficient AI models are allowing more language, vision, and reasoning workloads to run on compact edge hardware, according to Nvidia. The company is positioning the system for devices that need to process language and images and respond in real time.
The Jetson Orin Nano 2 delivers 78 trillion operations per second (TOPS) of AI compute and includes 8GB of memory and an eight-core Arm CPU. Nvidia said the system doubles the inference performance of the Jetson Orin Nano Super and increases AI and video-processing performance while retaining the same compact form factor.
The performance increase comes from upgraded Tensor Cores and higher memory bandwidth. Nvidia said the system uses 40% less power than its predecessor when delivering equivalent performance in 15-watt mode.
ARC Advisory Group identified machine vision as a major industrial edge AI application, with visual data processed on the factory floor for tasks including automated quality control and product sorting. The research firm cited latency, bandwidth, cost, security, and data privacy among the factors affecting cloud-based inference in industrial environments.
The Jetson Orin Nano 2 also supports large language models and vision-language models optimised for memory-efficient edge inference, including Nvidia Cosmos and Nemotron, Gemma 4, and Qwen 3.
More than three million developers are building applications on Nvidia’s robotics software stack, according to the company.
Machine vision moves to the edge
Industrial machine vision provides an existing example of Nvidia’s Jetson hardware being used alongside production equipment. Cognex launched its In-Sight 6900 Vision Controller in April 2026 using Nvidia Jetson technology to execute neural networks at the edge without an external PC or distributed computing architecture.
The controller provides up to 157 TOPS of AI performance and supports multiple high-resolution AI models operating simultaneously. Cognex said its Nvidia TensorRT integration is designed to keep AI inference synchronised with the microsecond-level timing of high-speed production lines.
The system is designed for inspection tasks where part sizes vary and defects differ in appearance. Cognex said its AI functions include classification and pixel-level segmentation, while cameras, optics, and lighting can be configured for specific inspection requirements.
A Cognex survey of more than 500 manufacturers, OEMs, and system integrators across North America, Europe, and Asia found that 57% of respondents were using AI in machine-vision operations. Another 30% said they planned to deploy it in the near term.
Cognex is also among the first companies identified as adopting or exploring the Jetson Orin Nano 2. The machine-vision supplier already uses Nvidia Jetson technology in its In-Sight 6900 controller, although neither company has disclosed how Cognex intends to use the new module.
Industrial computing supplier JWIPC has also outlined plans for the Orin Nano 2. The company plans to develop a JEA Series industrial vision edge computer using the module’s 78 TOPS of compute.
JWIPC said the fanless, fully enclosed system will operate at 15 watts and support feeds from two to four industrial cameras for real-time defect detection. The company said the design is intended as an alternative to industrial vision configurations that combine a larger PC with a separate GPU.
JWIPC described the JEA Series as a planned product and did not name customers in its announcement.
Advantech has also announced plans to expand its industrial edge systems with Orin Nano 2 support. The company identified intelligent machines, robotics, and other industrial applications among the areas targeted by the hardware.
AAEON, ADLINK, Aetina, Antmicro, Aptiv, Connect Tech, Seeed Studio, JWIPC, and other Jetson partners are also developing carrier boards, hardware systems, customised AI software, and reference designs around the module.
Edge AI moves into industrial robotics
JWIPC is also planning an Orin Nano 2-based controller for autonomous mobile robots and automated guided vehicles. The company said the controller will use an on-device architecture combining perception, decision-making, and control, with support for Nvidia Isaac and the Robot Operating System ecosystem.
The International Federation of Robotics (IFR) said AI-based mobile robots can combine data from cameras, LiDAR, and other sensors for simultaneous localisation and mapping, or SLAM, in warehouses and manufacturing environments.
The IFR also identified computer vision applications including object recognition, barcode reading, sorting, defect detection, quality inspection, and real-time production-line monitoring.
JWIPC said its planned robot controller will use the Orin Nano 2’s 15W operating mode. Nvidia said the module uses 40% less power than the Orin Nano Super when delivering equivalent performance in that mode.
NIST’s 2026 Roadmap on Artificial Intelligence and Machine Learning for Smart Manufacturing identifies industrial data management, integration with heterogeneous sensing and control systems, and trustworthy, explainable, and reliable operation among the challenges surrounding industrial AI. The roadmap also covers research areas including advanced sensing, autonomous systems, robotics, digital twins, industrial data analytics, reliability, and safety.
The Jetson Orin Nano 2 is not yet commercially available. Nvidia expects the production module and developer kit to become available in the first half of 2027.
(Photo by Nvidia)
See also: NVIDIA opens AV reasoning model to robotaxi developers
Want to learn more about IoT from industry leaders? Check out IoT Tech Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events, click here for more information.
Internet of Things News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
