NVIDIA Jetson AGX Thor Developer Kit – Ultra-High-Performance AI & Robotics Embedded System with 2070 FP4 TFLOPS Blackwell GPU, 128 GB RAM, Multi-Sensor & Vision AI Support – Edge AI Computer for Humanoid Robotics & Physical AI

The NVIDIA Jetson AGX Thor Developer Kit is the ultimate embedded computing platform for the next generation of physical AI, robotics, and autonomous systems. Built around the NVIDIA Blackwell GPU architecture, Jetson Thor delivers up to 2070 FP4 teraflops of AI compute in a configurable power envelope that scales from 40 W to 130 W, empowering developers to push the limits of edge AI performance. Engineered from the ground up for humanoid robotics and advanced AI inference, this embedded system replaces traditional compute constraints with high-performance, real-time reasoning capabilities. Its massive 128 GB of LPDDR5X memory and 273 GB/s memory bandwidth allow large generative AI models, vision-language action models, and multimodal sensor fusion to run efficiently at the edge without dependence on cloud servers. Ideal for robotics applications such as autonomous navigation, multi-sensor perception, real-time video analytics, and intelligent control systems, Jetson Thor integrates seamlessly with the NVIDIA Jetson software ecosystem. Access tools such as JetPack SDK, Isaac robotics frameworks, Holoscan sensor acceleration, and GR00T foundational AI models to build, simulate, and deploy complex AI-driven solutions faster than ever. Jetson Thor includes extensive I/O options, from multi-gigabit Ethernet, PCIe Gen5 slots, USB-C/A connectivity, to QSFP28 interfaces, making it easy to integrate with high-speed camera arrays, LIDAR, radar systems, and other robotic peripherals. Its developer-centric architecture ensures rapid iteration while maintaining robust performance and security for production-grade robots and embedded systems. From autonomous warehouses and industrial automation to healthcare robotics and next-gen research projects, Jetson Thor is the embedded AI supercomputer you can build on. Designed to deliver industry-leading throughput, scalable memory, and accelerated multi-sensor processing, it gives developers and innovators a powerful backbone for real-world AI and physical intelligence applications.