Advantech has introduced the MIC-743, a new Artificial Intelligence computing system built around the NVIDIA Jetson Thor module, Jetson T5000. The company positions the MIC-743 as a solution to bring server-grade Artificial Intelligence performance to edge deployments, targeting high-performance use cases. The system is described as ready for deployment in applications such as video language model (VLM) inferencing and large language models (LLMs).
The MIC-743 relies on the Jetson T5000 to deliver up to 2,070 FP4 TFLOPs of Artificial Intelligence performance, according to Advantech. Hardware specifications highlighted in the announcement include 2,560 CUDA cores, 96 Tensor cores and 128 GB of LPDDR5X memory. Advantech frames these resources as enabling real-time inferencing for complex edge workloads and providing the bandwidth and capacity required to run large models without bottlenecks.
Advantech says the platform can run multi-modal Artificial Intelligence tasks simultaneously, supporting VLM, LLM inferencing and intelligent video analytics at the edge. The company emphasizes the combination of compute, accelerator cores and memory as the rationale for deploying the MIC-743 in deep learning and computer vision scenarios where real-time performance and local processing are important. Availability details and pricing were not stated in the article.
