NVIDIA Unveils 865-TFLOP Jetson T3000 and T2000 for Edge AI, Eyes Q1 2027 Launch
Updated
Updated · NVIDIA Blog · Jul 16
NVIDIA Unveils 865-TFLOP Jetson T3000 and T2000 for Edge AI, Eyes Q1 2027 Launch
3 articles · Updated · NVIDIA Blog · Jul 16
Summary
NVIDIA introduced Jetson T3000 and T2000 Thor modules for mass-market robotics and edge AI, with commercial availability scheduled for Q1 2027 and T3000 emulation support arriving later this month in JetPack 7.2.1.
The T3000 delivers 865 FP4 teraflops in roughly half the size and power of the T5000, while the T2000 offers 400 FP4 teraflops and 16GB of memory as a lower-cost entry point.
NVIDIA said the modules target robots and autonomous machines running foundation models at the edge, with users including Amazon Robotics, Boston Dynamics, FANUC, Hitachi and 1X already building on Jetson AGX Thor.
The launch also adds Jetson agent skills that cut memory needs in days instead of weeks and a 4-billion-parameter Cosmos 3 Edge model for on-device robot reasoning and action generation.
With the new modules, NVIDIA says its Jetson lineup now spans 70 TOPS to 2,000 teraflops, broadening its push to make physical AI and humanoid systems deployable at scale.
Is NVIDIA's new Thor chip the final move in its plan to monopolize the entire AI industry, from cloud to robotics?
With NVIDIA's powerful robot chips not shipping until 2027, can any competitor disrupt its planned domination of physical AI?
NVIDIA Jetson T3000 and T2000: Blackwell-Powered Edge AI Modules Deliver Up to 865 Sparse FP4 TFLOPS for Scalable Robotics and Physical AI
Overview
NVIDIA has unveiled the Jetson T3000 and T2000 modules, which are part of the Jetson Thor family and built on the advanced Blackwell GPU architecture. These new modules bring high-performance AI to a wider range of edge applications, making it easier to deploy intelligent systems outside traditional data centers. The Jetson T3000 stands out with about 865 teraflops of compute power and 32GB of LPDDR5X RAM, offering strong performance for demanding edge AI tasks. Together, these modules are set to boost robotics and edge AI by providing powerful, efficient solutions for developers and manufacturers.