The three products on display are the MLA100, a PCIe card-type AI accelerator; the MLX-A1 Edge AI Box, an all-in-one edge AI system; and REGULUS, a single-board computer for on-device AI.

The MLA100 is a PCIe card-type AI accelerator powered by Mobilint's proprietary NPU, ARIES. Installed in existing servers or industrial PCs, it enables high-performance AI inference right out of the box, delivering 80 TOPS at just 25W and processing a wide range of workloads — from vision AI to large language models (LLMs) — in real time without cloud connectivity.

The MLX-A1 Edge AI Box is an all-in-one edge AI system with Mobilint's AI accelerator and software stack pre-installed. It runs diverse AI workloads — including LLMs, vision-language models (VLMs), video analytics, and speech recognition — in real time without a separate server or cloud connection. Since data never leaves the premises, it is well suited for enterprise and public-sector environments where data security and response speed are critical.

REGULUS is a standalone single-board computer that integrates a CPU, NPU, ISP, and multimedia codecs into a single SoC, delivering 10 TOPS at under 3W of power without a host system. It is optimized for on-device AI applications in the field, such as AI cameras, robots, and drones.

All three products come with a full-stack SDK compatible with major deep learning frameworks including TensorFlow, PyTorch, and ONNX, and support more than 490 open-source models, enabling customers to deploy their existing AI models with minimal development effort.

Through this exhibition, Mobilint aims to raise its brand recognition across the Asian market and expand its global partner network. Building on this foundation, the company plans to accelerate the supply of its AI semiconductor solutions across a wide range of industries, including smart factories, robotics, and drones.