Nvidia Launches Jetson Thor Chip, The New Brain of Next-Gen AI Robots

Nvidia has announced the release of Jetson Thor, a new robotics computer that is set to serve as the brain of smart machines. The chip promises more computing power than the older Jetson Orin and is designed to manage real-time tasks, such as robotic work that requires near-instantaneous data processing.

Jetson Thor provides 7.5 times the AI performance, 3.1 times faster CPU speeds, and double the memory of Jetson Orin, the company says. That will let robots rapidly digest information coming in from cameras, sensors, and other inputs directly on the device. Reducing reliance on the cloud allows robots to act more quickly and to operate in more complex environments.

Agility Robotics, the company that made the Digit humanoid robot, has already revealed that it will be utilizing Jetson Thor. Peggy Johnson, the company’s chief executive, said that the chip will allow Digit to respond better in real time and take on advanced tasks. Digit is already in use in warehouses and factories, in roles such as stacking and loading objects. Jetson Thor, the robot, should be able to tackle more demanding responsibilities.

Boston Dynamics, the company famous for robots like Atlas, will also adopt Jetson Thor. In a statement, Nvidia described the upgrade as server-level computing inside the robot. This will enable Atlas to be able to execute high-performance AI models and process huge amounts of data on board the device itself, minimizing latencies and making it more autonomous.

The possible applications for Jetson Thor doesn’t end with a humanoid robot; it might become a component in a surgical assistant, delivery robot, smart tractor, factory robotic arm, or an AI system that analyzes videos. Because it is designed for sophisticated AI models, such as large transformers and vision-language-action systems, it allows robots to interact with the world more intelligently.

Jetson Thor also enables integration with popular AI frameworks, including Cosmos Reason, DeepSeek, Llama, Gemini, and Qwen, as well as robotics models like Isaac GR00T. This enables developers to create robots capable of performing tasks that involve reasoning, planning, and vision.
Top-tier universities are also testing Jetson Thor. Teams at Stanford, Carnegie Mellon, and the University of Zurich are using it to enhance navigation, perceptual, and decision-making skills in robots. Sebastian Scherer, associate research professor at Carnegie Mellon, said the chip would let robots manage “much more nuanced tasks,” such as medical triage or search-and-rescue operations.

Nvidia has priced the Jetson Thor developer kit at $3,499. Jetson T5000 production modules will also be available to bulk buyers for $2,999 each. The company announced that its DRIVE AGX Thor kit for autonomous vehicles is now available for preorder, with shipment slated to begin in September.