NVIDIA Brings Greater Intelligence to Robotics with Jetson Thor

Get insight into NVIDIA’s Jetson AGX Thor, combining Blackwell GPUs and Arm CPUs to deliver high-efficiency AI performance for advanced robotic and embedded systems.

NVIDIA’s Jetson AGX Thor is the latest addition to the Jetson line of embedded processors, enhancing performance by leveraging the company’s newest-generation Blackwell architecture. The improved performance is suitable for high-end robotics and other demanding applications. Designed for what NVIDIA calls “Physical AI,” Thor aims to integrate motion, perception, and reasoning into a single module. At the top of the range, the Jetson T5000 pairs a Blackwell-generation GPU with an Arm Neoverse-V3AE CPU cluster to deliver market-leading AI performance within a 40-130 W envelope. The Jetson T4000 variant applies the same heterogeneous architecture to a 40-to-70-watt range to support mobile systems that prioritize energy efficiency over maximum throughput. In both variants, NVIDIA designed Thor to reduce the latency and contention that have traditionally separated perception, planning, and control workloads.