Thor-Hammer: Enormous Computing Power in Industrial Design

The Advantech AIMB-294 is a particularly powerful Mini-ITX mainboard based on the cutting-edge NVIDIA® Jetson Thor™ platform. Designed to meet complex requirements in AI, edge computing and image processing, it combines exceptional computing power with an industrial design. Featuring up to 2560 CUDA cores, 96 Tensor Cores, 128 GB LPDDR5X memory and AI performance of 2,070 TFLOPS (FP4), this board sets new standards in the field of embedded AI.

Thanks to its robust construction and innovative software tools, the AIMB-294 is ideal for applications such as medical image processing, AI-assisted diagnostics, robotics, intelligent quality control, and industrial automation.

Comprehensive Software and Developer Ecosystem

The AIMB-294 impresses not only with state-of-the-art hardware but, above all, with a powerful software platform that significantly accelerates and simplifies the development of AI solutions.

Easy Deployment with Embedded Software & Advantech Tools

With its complete embedded software ecosystem — including system images, configuration, and management tools — Advantech enables a fast and straightforward project start.

Highlights & Key Features

Massive AI and GPU Performance

• Powered by NVIDIA Jetson T5000 or T4000 modules with Arm® Neoverse®-V3AE CPUs.
• Up to 2560 Blackwell CUDA cores and 96 Tensor Cores for high-end deep learning workloads.
• Up to 2,070 TFLOPS (FP4) AI performance for highly complex inference-based applications.

High-Resolution Display Support

• 1× HDMI 2.1 (8K @ 30 Hz).
• 1× DisplayPort 1.4a (8K @ 30 Hz).

Flexible and Extensive I/O Configuration

• USB: 2× USB 3.2 Gen 2 (Rear), 1× USB-C, several internal USB headers.
• Network: QSFP28 up to 4× 25 GbE (depending on model) + 5 GbE RJ-45 internal.
• Serial & Fieldbus: 2× RS-232/485, 4× CAN (T5000 only).
• Expansion: M.2 M-Key (PCIe Gen5 x4 NVMe), M.2 B-Key (PCIe/USB/I²C), M.2 E-Key (WiFi/Bluetooth).
• Camera: GMSL Mini-FAKRA connectors for camera and sensor systems.

Linux Ubuntu & DeviceOn Remote Management

• Ubuntu Linux as a stable, widely used basis for AI and edge computing projects.
• DeviceOn enables professional remote management: OTA updates, health monitoring, remote debugging, and secure cloud access.

Advantech Edge AI SDK

• Includes runtime SDKs, model optimization tools and performance evaluation tools.
• Facilitates integration of neural networks, model conversion, optimization and inference acceleration.
• Ideal for projects requiring rapid development cycles and reproducible AI workflows.

Support for NVIDIA Clara & Holoscan

• NVIDIA Clara: For medical image analysis, AI-assisted diagnostics, and radiology workflows.
• NVIDIA Holoscan: For latency-critical edge AI applications – such as medical imaging, surgical assistance, and sensor fusion.