Nvidia Corporation has unveiled an updated version of its compact computing module, the Jetson Orin Nano 2, designed for robotics and edge artificial intelligence systems. The key feature of the new product is a twofold increase in inference performance while maintaining the same form factor, opening new horizons for running modern language and multimodal models directly on the device, without relying on cloud servers.
Technical breakthrough: twice as fast with lower power consumption
The Jetson Orin Nano 2 is based on an 8-core processor built on the Arm architecture, complemented by 8 GB of RAM. Peak AI computing performance reaches 78 TOPS. Thanks to improved tensor cores and increased memory bandwidth, inference efficiency has doubled compared to its predecessor, the Jetson Orin Nano Super.
Notably, Nvidia engineers managed to preserve the compact form factor of the previous version. In energy-efficient mode with power consumption of just 15 W, the module delivers performance on par with the previous model while consuming 40% less energy. This is critically important for autonomous devices—from robots and drones to computer vision systems—that must process data in real time, minimizing latency and dependence on cloud infrastructure.
Physical AI: reasoning and acting autonomously
Nvidia positions the new product as a platform for so-called "physical AI"—systems capable not just of analyzing, but also of perceiving the environment, making decisions, and interacting with it. The module supports advanced optimized models for edge inference, including NVIDIA Cosmos, NVIDIA Nemotron, Gemma 4, and Qwen 3. Developers gain tools for image and speech recognition, navigation, communication with humans, and autonomous decision-making.
An important trend I am tracking: modern compact models are already achieving accuracy comparable to the large systems of previous generations. This means we are witnessing a fundamental shift—generative AI is migrating en masse from data centers directly to edge devices. For the market, this means lower entry barriers and accelerated AI adoption in industry.
Ecosystem and early deployments
Nvidia's robotics stack is already used by more than 3 million developers. Among the first integrators of the Jetson Orin Nano 2 are machine vision system manufacturer Cognex, Doosan Bobcat equipment, and robotics company Matic, which applies the platform in home robots for simultaneous indoor mapping, object recognition, and interaction with people.
Special attention deserves the autonomous drone scenario: Alphabet's Wing division, which uses the previous generation of modules in its delivery systems, is already planning to test the new version to improve data processing speed and energy efficiency.
Nvidia's strategy is obvious—democratizing generative and agentic AI. The Jetson Orin Nano 2 occupies the lower segment of the lineup but offers computing capabilities that previously required significantly more powerful hardware. This is a direct response to the growing demand for small- and medium-scale autonomous systems.
My expert conclusion: the market is on the verge of a "ChatGPT moment" for robotics, as industry leaders put it. Hardware solutions like this are the catalyst for that process. By the end of 2027, we could see the mass emergence of affordable robots with embodied AI capable of operating without constant cloud support.