NVIDIA Introduces Jetson Thor T3000 and T2000 for Mass-Market Robotics and Edge AI

NVIDIA announced the Jetson Thor T3000 and T2000 modules to meet demand for compact, power-efficient AI supercomputers for general-purpose robots and autonomous machines. The T3000 delivers 865 FP4 teraflops of AI compute in a form factor half the size and power of the T5000, combining an NVIDIA Blackwell GPU, an eight-core Neoverse Arm CPU, 32GB LPDDR5X memory, and 273GB/s memory bandwidth. The T2000 offers 400 FP4 teraflops and 16GB memory for entry-level edge AI. Leading companies including 1X, Agile Robots, Amazon Robotics, Boston Dynamics, FANUC, Hitachi, and Techman Robot are building on the Jetson platform. NVIDIA also released new Jetson agent skills that automate memory optimization, allowing developers to achieve significant memory savings in days. Humanoid robotics leaders UBTech and Agile Robots, along with Connect Tech, reduced memory usage by up to 15GB. In smart retail, SandStar reduced memory by 4GB; GROOVE X optimized workload distribution for LOVOT; and NoTraffic reduced memory by 30% in intelligent transportation. Additionally, NVIDIA expanded its Cosmos 3 family with Cosmos 3 Edge, a 4-billion-parameter model for embodied systems that can be post-trained in about a day and deployed on Jetson Thor for real-time vision analysis. Developers can start using T3000 emulation mode with JetPack 7.2.1 later this month, with T2000 emulation support following later. The T3000 and T2000 modules will be available in Q1 2027.