Innodisk showed up strong at COMPUTEX 2025 with a booth full of industrial AI solutions that are anything but ordinary. Known for its deep roots in embedded systems and rugged tech, Innodisk is pushing further into AI computing, private LLM platforms, and next-gen edge vision systems. This year’s lineup shows what happens when real R&D meets real-world industrial use cases.
Qualcomm-Powered Smart Parking with Local AI Inference
Innodisk also brought out a smart parking and security solution powered by the Qualcomm Dragonwing IQ9 platform. Built with an integrated Neural Processing Unit (NPU), the system can handle advanced AI workloads like vision-language models and behavior tracking, all running at the edge with no need for cloud processing.
With fully local processing and support for expandable camera modules, this setup is tailor-made for smart city environments where low-latency responses and real-time decision-making are essential.
Intel Core Ultra Reference Kit and MIPI-over-Type C Camera
Partnering with Intel, Innodisk also unveiled a Core Ultra Series 2 reference kit that includes a range of its own components: MIPI-over-Type C camera modules, DDR5 memory, SSDs, and additional I/O options. The idea is to help integrators and developers quickly build Intel-powered edge systems with plug-and-play compatibility.
Next-Gen Storage and Memory for the AI Era
On the storage side, Innodisk showed its latest PCIe Gen5 SSDs, available in U.2 and OCP-compliant EDSFF formats like E3.L, E3.S, and E1.S. These drives are tuned for enterprise servers, AI data centers, and high-speed edge platforms where Gen5 performance really makes a difference.
They also had new LPDDR5X CAMM2 and DDR5 MRDIMM modules on display, giving server builders and OEMs more memory options for high-bandwidth, low-latency workloads.
APEX-X100 and AccelBrain: Private LLM Without the Cloud
One of the standout highlights this year is the APEX-X100, an AI computing platform built for on-premise LLM training and inference. It runs fully local using Innodisk’s AccelBrain software, allowing businesses to train, fine-tune, and deploy large language models without relying on cloud services.
Perfect for privacy-conscious industries like finance, manufacturing, or logistics, the APEX-X100 is a compact but powerful system tailored for enterprise-grade AI tasks running entirely within local infrastructure.
APEX-S100: Short Depth, Full GPU Firepower
If you need GPU muscle but don’t have space for a massive rack, Innodisk’s APEX-S100 short-depth server might be the solution. It fits into a 2U, 420mm-depth chassis yet supports up to two double-width GPUs, along with multiple memory expansion slots.
This server is ideal for large-scale analytics and AI inference workloads in environments where rack space is limited. It’s small, but it doesn’t compromise on throughput or expansion.
Final Thoughts
Innodisk’s COMPUTEX 2025 lineup feels like a blueprint for next-gen industrial AI systems. From private LLM deployments and GPU-ready short-depth servers, to modular vision systems and cloud-free edge AI, everything on display is geared for practical, scalable deployment. Add in Gen5 storage and high-performance DDR5, and it’s clear that Innodisk is not just riding the AI wave, they’re helping shape how it’s built from the ground up.
