Advantech demos Jetson Thor AI at GTC Paris 2025

The booth will feature a live demo of video analytics AI agents powered by the NVIDIA AI Blueprint for video Search and Summarization (VSS), accelerated by NVIDIA Jetson Thor, a showcase of the MIC-736 with dual NVIDIA Jetson AGX Orin modules, and the NVIDIA MGX server platform designed for large-scale AI model training and simulation workloads.

Live demo: high-performance video AI at the Edge with NVIDIA Jetson Thor

Visitors will experience a live demonstration of Advantech’s MIC-743, an Edge-ready platform built on the latest NVIDIA Jetson Thor, purpose-built for high-throughput AI workloads. The demo showcases the VSS blueprint uses NVIDIA Metropolis services to analyse video streams in real-time. In addition to running on server-class NVIDIA accelerated computing, the VSS blueprint can now run entirely on a compact Edge system. The demo highlights full-frame video ingestion, semantic search, and rapid inference, all executed fully at the Edge without relying on Cloud computing.

MIC-743 key features:

• Up to 2,070 FP4 TFLOPS AI performance and 128GB LPDDR5X memory
• 100GbE QSFP28 high-bandwidth connectivity for sensor input
• Rugged, compact design supporting extended temperature and vibration tolerance
• Rich industrial I/O and GMSL2 camera support

MIC-743 is ideal for robotics, automated inspection, mobility, and AI-based applications where real-time response and Edge processing are mission-critical.

Demo: dual AGX Orin-powered MIC-736 for scalable inference

Also on display is MIC-736, Advantech’s high-performance, fan-less Edge AI system built with two NVIDIA Jetson AGX Orin modules. Designed for compute-intensive AI inference and multi-stream vision processing, MIC-736 delivers robust performance for industrial automation, transportation, and safety systems.

MIC-736 key features:

• Dual Jetson AGX Orin modules for powerful Edge inference
• Multiple GbE, HDMI, GMSL2, and CANBus ports for flexible I/O
• Fanless, wide-temp design for harsh industrial environments
• Optimised for robotics, smart inspection, and embedded video analytics

Demo: SKY-622G4MGX GPU server for omniverse & AI scale-up

Advantech will also showcase a static demo of its MGX server platform with the NVIDIA RTX Pro 6000 Blackwell Server Edition GPU, built to support AI model simulation and training at scale. Leveraging NVIDIA MGX architecture, the server is optimised for hybrid AI workloads across both Cloud and on-premise environments, helping developers accelerate simulate-train-inference cycles while maintaining deployment flexibility.

Full-stack AI enablement: simulate, train, infer – all at the Edge beyond hardware, Advantech offers a complete AI enablement stack – from simulation and model training to deployment and real-time inference. Advantech helps customers accelerate development, simplify integration, and scale AI across diverse use cases.

Advantech helps customers:

• Integrate Holoscan platform: real-time sensor processing for low-latency AI applications
• Seamless workflow: from model simulation to Edge deployment
• AI-ready platforms: optimised for VLMs, robotics, and video analytics

Designed for industrial-grade reliability and scalability, Advantech's Edge AI platforms power cutting-edge applications in machine vision, smart manufacturing, robotics, and real-time video analytics. Whether you're deploying perception AI, Generative AI, agentic AI, or physical AI, Advantech solutions provide the performance, flexibility, and validated software-hardware integration required to accelerate innovation at the Edge.

Experience Advantech at GTC Paris 2025

Join Advantech at NVIDIA GTC Paris 2025 to see how NVIDIA Jetson Thor, Dual Jetson Orin, and MGX platforms are transforming the Edge AI landscape. Discover how Advantech empowers real-time AI for robotics, vision analytics, and scalable training with rugged, reliable, and ready-to-deploy solutions.