Nvidia unveiled its new Jetson Thor T3000 and T2000 modules today, compact AI supercomputers designed to run foundation models directly on autonomous machines and general-purpose robots. The modules are built on Nvidia's Thor architecture and promise to bring data center-class AI performance to the edge. The announcement targets the shift of general-purpose robots from controlled lab environments to factory floors, warehouses, and potentially homes, creating demand for compute that can handle complex AI workloads in compact packages. Nvidia has not released detailed specs yet, but the T3000 and T2000 designations suggest a tiered approach to hit different price and performance points. The focus on foundation models at the edge enables lower latency, better privacy, and robots that can operate without constant internet connectivity. Nvidia's dominance in AI accelerators gives it a natural advantage, as the company already powers the training infrastructure for most major AI models, allowing developers to train on Nvidia GPUs in the cloud and deploy on Nvidia modules in robots.