NVIDIA has introduced its next-generation AI platform, designed to bring artificial intelligence out of the digital realm and into the physical world. Called IGX Thor, this powerful industrial-grade processor aims to enhance real-time perception, reasoning, and decision-making for machines operating in factories, hospitals, and autonomous systems. The platform is expected to form the backbone of smart, safe, and responsive systems that can work seamlessly alongside humans.
NVIDIA IGX Thor introduced!
Compared to its predecessor, IGX Orin, the new IGX Thor delivers up to eight times higher AI computing performance. At its core lies NVIDIA’s latest Blackwell GPU architecture, which combines both integrated and discrete GPUs. The system provides an impressive 5,581 FP4 TeraFLOPS of computing power and supports 400 GbE connectivity, enabling rapid data transfer and analysis directly at the device level.
One of IGX Thor’s most significant advancements is its ability to run AI models directly on the device, eliminating latency and enabling instant responses. This capability allows industrial robots to detect faults within milliseconds or surgical robots to process high-resolution images in real time during procedures. NVIDIA also emphasized that IGX Thor is engineered for longevity, featuring a 10-year life cycle and long-term software support for demanding industrial environments.
The platform runs the NVIDIA AI Enterprise software suite, which includes key components such as Isaac for robotics, Metropolis for visual AI, Holoscan for sensor data processing, and NIM for microservice-based AI development. IGX Thor also integrates NVIDIA’s Halos safety system to ensure functional safety and secure collaboration between humans and machines in robotic and medical applications.
Several global technology leaders have already started integrating IGX Thor into their operations. Hitachi Rail is deploying the platform across its rail networks for predictive maintenance and autonomous inspection. Maven Robotics is utilizing it as the central processor for its next-generation industrial robots, while CMR Surgical is testing IGX Thor for AI-assisted surgical systems capable of real-time analysis and adaptive decision-making in the operating room.
According to CMR Surgical’s Chief Technology Officer, the system’s real-time data processing capabilities enable safer, more precise surgical procedures, allowing doctors to perform complex operations with greater accuracy and reduced risk.
The IGX Thor ecosystem includes major hardware partners such as Advantech, ADLINK, ASRock Rack, Inventec, NexCOBOT, Onyx, and WOLF Advanced Technology. These companies are developing industrial servers, mainboards, camera systems, and sensors optimized for the new platform.
NVIDIA announced that IGX Thor will be available in two configurations: the IGX T5000 module and the IGX T7000 board kit. Both are designed for certified, safety-compliant physical AI applications. Developer kits and production versions are expected to launch in December.
With IGX Thor, NVIDIA is taking a major step toward bringing artificial intelligence from cloud data centers to real-world environments. The platform represents a pivotal moment in the evolution of industrial and medical AI, marking the beginning of a new era where machines can think, sense, and act in real time.
