Delivering powerful performance for VLM and LLM applications at the edge

Advantech has launched the MIC-743, its latest AI computing system powered by the NVIDIA Jetson Thor. Designed to bring server-grade AI performance to the edge, it is now ready for deployment in high-performance applications such as VLM and LLMs.

Equipped with the NVIDIA Jetson Thor module, Jetson T5000, the system delivers up to 2,070 FP4 TFLOPs of AI performance, allowing real-time inferencing for complex edge applications. The 2560 CUDA cores and 96 Tensor cores provide massive processing power for AI workloads such as deep learning and computer vision. The 128GB LPDDR5X memory offers sufficient bandwidth and capacity to handle large AI models, including VLM and LLM inferencing, without bottlenecks. Together, these features empower the system to run multi-modal AI tasks simultaneously, supporting VLM, large language models, and intelligent video analytics at the edge.

With its compact, industrial-grade design and rich I/O connectivity, MIC-743 is ideal for deployment in diverse environments such as factories, research labs, and development centres. It also allows seamless integration with various sensors, cameras, and peripheral devices to build complete AI solutions quickly. This flexibility will enable developers and system integrators to accelerate their projects, decrease deployment time, and unlock new possibilities in intelligent edge applications.

Visit the company at Taipei Automation 2025, from 20-23 August 2025, where it will showcase diverse and flexible AI solutions, including MIC-743, powering edge VLM, delivering a stable foundation for various front-end applications.