NVIDIA IGX Thor

IGX is production-ready, combining industrial-grade hardware with enterprise software and support.

The NVIDIA IGX Thor Platform unlocks real-time sensor processing and AI reasoning for physical AI agents with industrial-grade hardware, enterprise software, and functional safety. It’s powered by the [NVIDIA Blackwell Architecture](https://www.nvidia.com/en-in/data-center/technologies/blackwell-architecture/) iGPU and an optional dGPU, delivering up to 5581 FP4 TFLOPS of AI compute to effortlessly run multiple generative AI models at the edge. Compared to NVIDIA IGX Orin™, it provides up to 8x higher AI compute on iGPU, 2.5x higher AI compute on dGPU, and 2x better connectivity.

NVIDIA IGX Thor unifies high‑performance AI compute with functional safety for robots operating in dynamic environments. It supports both “inside‑out” safety using onboard sensors and “outside‑in” safety using infrastructure sensors. At its heart is a dedicated Functional Safety Island that provides an independent safety processor that isolates safety‑critical workloads — designed to meet ISO 26262 and IEC 61508 at ASIL D/SC3 and ASIL/SIL 2 levels.

NVIDIA AI Enterprise software powers edge AI on NVIDIA IGX Thor — ensuring mission-critical performance, security, and long-term enterprise support across the entire stack. It seamlessly runs generative AI — from VLA models like NVIDIA Isaac GR00T N to popular LLMs and VLMs such as NVIDIA Cosmos Reason — and brings cloud-native NVIDIA AI Enterprise and NVIDIA NIM™ to the edge. Applications include NVIDIA Isaac™ for robotics, Metropolis for visual AI, Holoscan for sensor processing, and NVIDIA Halos safety AI agents. This gives your organization a secure, stable, and high-performance experience for next-gen physical AI.

NVIDIA IGX Thor NVIDIA IGX T7000 The MicroATX board kit with dGPU support, functional safety, NVIDIA® ConnectX®-7, and BMC support. NVIDIA IGX T5000 The IGX Thor System-on-Module for custom carrier boards solution with optional functional safety features. NVIDIA IGX Thor Developer Kit The development kit with pre-installed dGPU for all IGX Thor use cases. NVIDIA IGX Thor Developer Kit Mini The development kit for robotic safety. NVIDIA IGX Orin NVIDIA IGX Orin 700 The MicroATX board kit with dGPU support, functional safety, NVIDIA ConnectX-7, and BMC support. NVIDIA IGX Orin Developer Kit The development kit for all IGX Orin use cases.

IGX Thor Developer Kit dGPU: With NVIDIA RTX PRO™ 6000 Blackwell Max-Q Workstation Edition AI Performance: Up to 5581 TFLOPS (FP4-Sparse) iGPU: 2560-core NVIDIA Blackwell architecture GPU with fifth-gen Tensor Cores Multi-Instance GPU (MIG) with 10 TPCs iGPU Max Frequency: 1.57 GHz dGPU: 24,064-core NVIDIA Blackwell architecture GPU with fifth-gen Tensor Cores CPU: 14-core Arm® Neoverse®-V3AE 64-bit CPU 64 KB I-Cache, 64 KB D-Cache 1 MB L2 Cache per core 16 MB Shared System L3 Cache CPU Max Frequency: 2.6 GHz Vision Accelerator: 1x PVA v3 Memory: 128 GB 256-bit LPDDR5X | 273 GB/s<br> 96 GB GDDR7 dGPU memory | 1792 GB/s Storage: 1 TB M.2 NVMe (PCIE Gen5 x2) Video Encode: 2x NVEncode (iGPU) 4x NVEncode (dGPU) Video Decode: 2x NVDecode (iGPU) 4x NVDecode (dGPU)x PCIe*: 2x Gen5 PCIe (x8, x16) USB*: USB 3.2 Gen2 Type-A stacked connectors<br> USB 3.2 Gen2 Type-C connector Networking*: 2x RJ45 (1GbE each)<br> 2x QSFP112 (200GbE Each)<br> Supports ConnectX-7 ConnectX Support: Yes Display: 1x VESA DisplayPort 1.4a Other I/O: 1x Line-Out and 1x MIC (Audio) BMC Support: Yes Safety Support: Functional Safety Island on SoC  Safety MCU on the carrier board Power: 40 W–130 W TMP<br> up to 300 W for dGPU Mechanical: 382.7 mm x 262.7 mm x 151.2 mm IGX Thor Developer Kit Mini dGPU: – AI Performance: 2070 TFLOPS (FP4-Sparse) iGPU: 2560-core NVIDIA Blackwell architecture GPU with fifth-gen Tensor Cores Multi-Instance GPU (MIG) with 10 TPCs iGPU Max Frequency: 1.57 GHz dGPU: – CPU: 14-core Arm® Neoverse®-V3AE 64-bit CPU 64 KB I-Cache, 64 KB D-Cache 1 MB L2 Cache per core 16 MB Shared System L3 Cache CPU Max Frequency: 2.6 GHz Vision Accelerator: 1x PVA v3 Memory: 128 GB 256-bit LPDDR5X 273 GB/s Storage: Primary Storage - 1TB M.2 M NVME (PCIe x4 Gen5) Secondary Storage: M2.M UFS3.1 Video Encode: 2x NVEncode Video Decode: 2x NVDecode PCIe*: M.2 Key M slot with x4 PCIe Gen5 (populated with 1 TB NVMe) M.2 Key E slot with x1 PCIe Gen5 (populated with WiFi 6e + Bluetooth Module) USB*: 2x USB Type-A 3.2 Gen2<br> 2x USB Type-C 3.1<br> 1x USB Type-C (Debug purpose only) Networking*: 1x 5GBe RJ45 connector<br> 1x QSFP28 (4x 25GbE)<br> Wi-Fi 6E (Populated on M.2 Key E slot<br> with x1 PCIe Gen5) ConnectX Support: No Display: 1x VESA DisplayPort 1.4a Other I/O: 2x FSI CAN Header<br> 2x 6-pin automation header<br> 2x 5-pin header JTAG Connector<br> 1x 4-pin Fan Connector: 12 V, PWM, and Tach<br> 2-pin RTC backup battery connector Microfit power jack<br> Power, Force Recovery, and Reset buttons BMC Support: No Safety Support: Functional Safety Island on the SoC Safety MCU on the carrier board Power: 40 W–130 W Mechanical: 243.19 mm x 112.40 mm x 56.88 mm (height includes feet, carrier board, module, and thermal solution)