Nvidia has begun selling its Jetson AGX Thor developer kit, a compact computing module it calls a “robot brain,” for $3,499. The product, announced on 25 August, is shipping now, while Jetson T5000 production boards will follow next month at $2,999 apiece for orders of at least 1,000 units. Built on the company’s new Blackwell GPU architecture, Jetson Thor delivers up to 2,070 FP4 teraflops of performance—about 7.5 times the AI compute and 3.5 times the energy efficiency of the prior-generation Jetson Orin—within a 130-watt power envelope. The module pairs 128 GB of memory with 14 Arm CPU cores, allowing robots to run large language, vision and multimodal generative AI models directly on-device rather than in the cloud. Amazon Robotics, Meta, Agility Robotics, Boston Dynamics and Caterpillar are among the early adopters integrating the hardware into warehouse, humanoid and industrial robots. Nvidia is also offering a Drive AGX Thor kit, now on preorder, for autonomous-vehicle developers. “We do not build robots, but we enable the whole industry with our infrastructure computers and the associated software,” said Deepu Talla, vice president of robotics and edge AI. While robotics still accounts for roughly 1 % of Nvidia’s revenue, the company regards the sector—alongside AI data-center chips—as a key long-term growth driver.