Innodisk Unveils AI and Embedded Vision Platform at COMPUTEX 2025

At COMPUTEX 2025, Innodisk presented a comprehensive range of AI computing and embedded vision systems tailored for industrial and enterprise use. The company’s showcase emphasised platform compatibility and AI acceleration at the edge—underlining its ambition to serve next-generation data-intensive applications, from smart cities to autonomous fleets.

Edge AI Infrastructure Takes Centre Stage

The focal point of Innodisk’s COMPUTEX presence was its APEX-X100 platform—designed for on-premise private LLM (large language model) applications. Developed to support local AI training, it integrates with AccelBrain, a software tool enabling scalable and efficient AI processing without reliance on cloud infrastructure. This is complemented by the APEX-S100, a compact 2U short-depth server optimised for high-performance edge workloads with support for dual double-width GPUs and multiple memory configurations.

Expanded Camera Module Lineup for Machine Vision

Recognising the critical role of visual data in industrial AI, Innodisk introduced an expanded portfolio of embedded camera modules supporting USB, MIPI, and GMSL2 protocols. Designed for compatibility with systems ranging from Raspberry Pi to AI servers, these modules are backed by dedicated engineering support and custom drivers.

In collaboration with Advantech, Innodisk demonstrated 16-channel GMSL2 camera integration on rugged edge platforms like the TREK-50N and REX-52N. These solutions target heavy-duty vehicle monitoring, offering synchronised multi-angle video capture and real-time driver analysis.

AI at the Edge with Qualcomm and Intel Platforms

Innodisk also highlighted its partnership with Qualcomm Technologies, showcasing a smart parking and security solution based on the Qualcomm Dragonwing™ IQ9 platform. Featuring an onboard Neural Processing Unit (NPU), the system supports real-time inference for vehicle recognition and behaviour tracking—capabilities critical to smart city infrastructure where latency and data privacy are priorities.

The company further introduced a reference kit built on Intel® Core™ Ultra Series 2, featuring Innodisk’s own MIPI-over-Type C camera modules, DRAM, and SSDs. This kit supports developers and OEMs building AI-driven systems on Intel’s latest platform.

Storage and Memory for AI Workloads

On the memory front, Innodisk displayed its LPDDR5X CAMM2 and DDR5 MRDIMM modules, offering high-speed, low-latency performance optimised for AI servers. These are accompanied by PCIe Gen5 SSDs in U.2 and EDSFF (E3.L, E3.S, and E1.S) formats, fully compliant with OCP SSD 2.0 specifications. These solutions are targeted at data centres and enterprise environments seeking high-throughput, low-power storage systems.

‍

These technologies support use cases like real-time fleet monitoring, autonomous inspection, and infrastructure intelligence—areas where tyre companies and commercial operators are increasingly active. As vehicle systems become more sensor-driven and localised AI becomes essential, Innodisk’s portfolio reflects broader movement towards decentralised, embedded intelligence.

‍

‍

Tagged with: edge AI, embedded vision, smart city infrastructure, AI computing platform, private LLM, GMSL2 camera, PCIe Gen5 SSD, LPDDR5X CAMM2, industrial AI, Innodisk COMPUTEX 2025

‍

‍