NVIDIA has launched the IGX Thor family of platforms designed to power industrial, medical, and robotics applications at the edge. As sectors increasingly adopt complex generative AI models, high-fidelity sensors, and demanding workloads, there is a critical need for hardware that balances massive performance with strict functional safety and long-term reliability. The IGX Thor series addresses these challenges by integrating enterprise-grade computing with industrial-grade durability. The platform family consists of four distinct configurations: the IGX T5000 System-on-Module, the IGX T7000 Board Kit, the IGX Thor Developer Kit, and the IGX Thor Developer Kit Mini. These systems are built around the Blackwell GPU architecture and 14-core Arm Neoverse-V3AE processors. The IGX T5000 SoM delivers up to 2,070 FP4 TFLOPS of AI performance in a compact form factor, suitable for custom carrier boards. The IGX T7000 Board Kit offers higher scalability, combining an integrated GPU with a discrete RTX PRO 6000 Blackwell Max-Q workstation edition GPU to achieve up to 5,581 FP4 TFLOPS. This configuration enables the system to handle significantly more interactive users and concurrent generative AI tasks compared to previous solutions like the NVIDIA IGX Orin. Performance metrics highlight a substantial leap in capability. The IGX T7000 provides up to eight times higher AI compute on its integrated GPU and 2.5 times higher performance with discrete acceleration. In real-world testing with large language models, the IGX T7000 demonstrated speedups ranging from 1.1x to 4.9x over the IGX Orin 700, while supporting a much higher number of concurrent users. Furthermore, the platform features dual 200 Gigabit Ethernet interfaces powered by NVIDIA ConnectX-7 SmartNICs. This high-speed connectivity utilizes RDMA technology to allow sensor data to flow directly into GPU memory, bypassing the CPU. This architecture is essential for real-time sensor fusion in environments requiring deterministic latency, such as autonomous mobile robots and surgical robotics. Safety and compliance are central to the IGX Thor design. The platforms are engineered to meet rigorous standards including ISO 26262 and IEC 61508, targeting ASIL D and SIL 2 certification levels. Each unit includes a Functional Safety Island (FSI) within the system-on-chip and, in select models, a dedicated safety MCU. The hardware is also built to withstand harsh physical conditions, including extreme temperatures, shock, and vibration, supported by extended temperature ranges and error-correcting code memory. NVIDIA has committed to a 10-year lifecycle for the IGX platform, ensuring long-term supply chain assurance, security updates, and stable software support. The IGX AI Enterprise software stack provides version-locked frameworks and SDKs maintained for a decade, alongside enterprise service level agreements. This ensures that deployments remain secure and compliant throughout their operational life. Transitioning from the previous Jetson Thor platform is seamless, as the IGX T5000 maintains pin and form factor compatibility, allowing for easy migration while benefiting from enhanced industrial robustness. The IGX Thor Developer Kit and Developer Kit Mini are now available for purchase through global distributors. The platform is positioned to transform industries ranging from smart manufacturing and healthcare to energy and infrastructure by enabling the deployment of advanced AI in safety-critical, regulated environments.