Advantech explained its edge AI and physical AI innovations at NVIDIA GTC 2026. Source: Advantech

At NVIDIA GTC this week, Advantech Co. is showing next-generation edge AI platforms and solutions powered by technologies including NVIDIA Jetson Thor and NVIDIA IGX Thor. By integrating hardware platforms, AI development frameworks, and ecosystem partnerships, Advantech said it is helping enterprises move efficiently from AI evaluation to scalable deployment at the edge.

As AI rapidly evolves, industry is moving beyond cloud-based model training toward real-world, physical applications—often referred to as “physical AI,” noted Ween Niu, general manager of Advantech North America. The Taipei, Taiwan-based company has long focused on edge computing and edge AI, building industrial-grade platforms designed for performance, reliability, and scalable deployment, he said.

At this year’s GTC, Advantech is demonstrating its close collaboration within the NVIDIA ecosystem, combining advanced hardware, software frameworks, and industry-ready solutions. The IoT and embedded systems company said they will accelerate physical and edge AI adoption in robotics, medical devices, logistics, and retail environments.

Advantech supports robotic perception, humanoid development

For robotics applications, Advantech is presenting edge AI platforms designed for rapid integration of sensors and multi-camera systems. The ASR-A702 and AFE-A702 are built on NVIDIA Jetson Thor. They work with Advantech Robotic Suite and NVIDIA Isaac ROS to support key perception capabilities such as object detection, distance estimation, pose tracking, and VSLAM (visual simultaneous localization and mapping).

To support the next generation of humanoid robotics, the MIC-742, also built on NVIDIA Jetson Thor, delivers up to 2,070 TFLOPS (FP4) AI performance. It can be paired with the NVIDIA Holoscan Sensor Bridge to enable an ultra-low-latency sensor-to-inference pipeline for transformer and vision-language-action (VLA) models. Advantach said these capabilities can accelerate robotics innovation and deployment.

The company added that it participates in the NVIDIA Halos AI Systems Inspection Lab and has partnered with FORT Robotics to integrate their embedded safety architecture. It designed the MIC-735, powered by the NVIDIA IGX T5000 module, to combine high-speed sensor processing, enterprise-grade reliability, and functional safety for autonomous mobile robots (AMRs) and other robotics deployments.

Advantech has collaborated with NVIDIA and FORT Robotics to build safety functionality into AMRs. Source: Advantech

Real-time medical imaging enables intelligent surgical systems

For healthcare, Advantech is exhibiting at GTC the AIMB-294 medical AI board, built on NVIDIA Jetson Thor. It can perform real-time surgical instrument anomaly detection, organ segmentation, and augmented reality (AR) overlay while operating at only 130W and without additional GPU modules.

Combined with advanced medical AI frameworks and edge AI software tools, the AIMB-294 supports low-latency imaging pipelines and streamlined AI model deployment. This will improve overall device performance and reliability, claimed Advantech.

The company also highlighted the USM-500, a medical-grade platform powered by NVIDIA IGX optimized for advanced AI-powered medical applications. It said the platform enables real-time multimodal sensor fusion and edge AI processing for AI-assisted surgery, intraoperative imaging, endoscopic video analytics, and robotic surgical guidance, delivering hospital-grade reliability and performance.

Advantech yesterday said it has expanded its medical portfolio with NVIDIA IGX Thor, including the MIC-735 industrial AI inference system, the AIR-427A medical-grade edge AI HPC, and the USM-500.

Advantech has released medical-grade processing using NVIDIA technology. Source: Advantech

Advantech designs visual AI agents, video intelligence for smart warehouses

To address the growing demand for intelligent logistics and warehouse management, Advantech is presenting the AIR-075, built on NVIDIA Jetson Thor. The visual AI platform is designed to enable visual AI agents for safety monitoring and incident response, with 10GbE connectivity supporting high-bandwidth video streaming.

Advantech will also demonstrate the MIC-743 running advanced video search and summarization capabilities that enable real-time vision analytics, natural-language queries, and rapid video summarization. These technologies, built on Jetson Thor, can help enterprises improve operational visibility, efficiency, and decision-making across logistics environments, said the company.

Edge-based generative AI powers smart retail experiences

For retail and service environments, Advantech is featuring the DS-015 edge AI system, built on NVIDIA Jetson Orin, running generative AI models through Advantech Edge AI SDK. It enables on-device large language model (LLM) interactions without relying on cloud connectivity, making it suitable for kiosks, smart retail stores, digital signage, and industrial edge deployments.

The company said its approach allows organizations to enhance customer engagement while maintaining low latency, data privacy, and operational reliability. In addition, Advantech last week announced its collaboration with Qualcomm to launch the SKY-641E3 server powered by the Qualcomm Dragonwing for on-prem generative AI.

Advantech North America invited industry partners and customers to visit Booth 1134 and Meeting Rooms 6077 and 6078 at NVIDIA GTC 2026 to explore how edge AI and physical AI technologies can accelerate digital transformation and unlock new opportunities for real-world applications.