NVIDIA IGX Thor: Enterprise Edge AI Platform
NVIDIA has announced IGX Thor, a comprehensive platform designed for deploying high-performance AI at the industrial and medical edge. The platform combines server-class compute with industrial-grade hardware, functional safety capabilities, and extended lifecycle support—essential for mission-critical edge deployments in factories, hospitals, and autonomous systems.
Platform Configuration
The IGX Thor family consists of four purpose-built configurations:
- NVIDIA IGX T5000 SoM: A compact, embedded module delivering up to 2,070 FP4 TFLOPS, designed for custom carrier boards and domain-specific industrial systems
- NVIDIA IGX T7000 Board Kit: The highest-performance variant with up to 5,581 FP4 TFLOPS on a MicroATX form factor, supporting discrete GPU expansion and dual 200 GbE networking
- NVIDIA IGX Thor Developer Kit: A full-featured development platform with advanced sensing and robotics support
- NVIDIA IGX Thor Developer Kit Mini: A space and power-constrained variant ideal for mobile robots and compact systems
Key Technical Features
All variants feature NVIDIA Blackwell GPU architecture with 2,560-core iGPU and 14-core Arm Neoverse-V3AE CPU. Memory configurations include 256-bit 128 GB LPDDR5x with 273 GB/s bandwidth. Networking capabilities include up to 4x 25 Gbps MGBE ports on the T5000 and dual 200 GbE QSFP112 ports on the T7000 for low-latency sensor data handling and RDMA support.
Safety and Compliance
The platform incorporates Functional Safety Islands (FSI) integrated into the SoC, with additional safety MCUs on higher-tier models. This architecture enables compliance with ISO 26262 (automotive functional safety) and IEC 61508 standards, critical for regulated industrial and medical environments. The platform supports real-time Linux for deterministic behavior and verifiable functional safety.
Enterprise Support
IGX Thor includes extended lifecycle support and enterprise-grade software stacks, providing long-term hardware and software maintenance essential for mission-critical deployments. The platform is designed for harsh physical conditions, secure infrastructure integration, and state-of-the-art AI inference and sensor fusion pipelines at the edge.