 |
Collaborating with Leading Global Companies to Showcase Heterogeneous AI Platform Deployments in Industrial and Enterprise Scenarios
SHENZHENMay 27, 2026 /PRNewswire/ — Innodisk, a global leader in edge AI solutions and industrial-grade storage, will showcase its full-stack ecosystem solutions at COMPUTEX 2026. The exhibition covers the five core layers of edge AI architecture: computing, memory, storage, sensing and communication, and software. It not only offers a comprehensive product lineup compatible with next-generation computing specifications but also, through the integration of modules and platforms, creates typical application scenarios such as enterprise-side local large language model (LLM) fine-tuning, intelligent security for heavy machinery, autonomous mobile robots (AMRs), and OCR container identification. This fully validates the large-scale deployment capabilities of edge AI.
Innodisk at COMPUTEX 2026, Building the Five Key Layers of Edge AI Architecture
Innodisk Group Chairman Chuan-Sheng Chien stated, “In 2025, we proposed a dual-driven market strategy of ‘Industrial AI’ and ‘Enterprise AI.’ At this year’s COMPUTEX, we are presenting tangible results in collaboration with international partners providing advanced computing architectures and vertical domain experts. Through our edge AI solutions covering the ‘five-layer architecture,’ we have successfully transformed AI computing power into solutions that can be genuinely applied to real-world industrial and enterprise scenarios, helping global customers accelerate their intelligent upgrades within a secure, reliable, and scalable framework.”
Focus on Enterprise AI: Local LLM Solutions Ensuring Data Sovereignty and Security
A major highlight of this year’s exhibition is Innodisk’s locally deployed AI solutions, designed to enhance the security and convenience of enterprise AI deployment. Among them, AccelBrain, based on the APEX-X200 edge AI platform, enables local LLM deployment with fully controlled data sovereignty. Complementing this, AccelTune offers a no-code LLM fine-tuning experience and can run on the APEX-S100 edge AI server, equipped with Intel® Xeon® 6700 processors and supporting dual NVIDIA RTX PRO GPUs. Used together, these two solutions help enterprises build secure LLM workflows tailored to specific business needs without relying on public clouds.
Industrial AI Edge Application Examples: Intelligent Industrial Safety Protection in Harsh Environments
In addition to enterprise applications, Innodisk will also demonstrate how edge AI delivers value in harsh industrial environments. The exhibition will feature a heavy machinery safety solution, integrating eight Innodisk GMSL2 camera modules with IP67 and IP69K protection ratings. This provides AI-driven panoramic image stitching, driver monitoring systems (DMS), and blind spot detection (BSD), enhancing operators’ environmental awareness and operational safety in outdoor construction scenarios.
As an Intel Gold Partner, Innodisk will also launch the APEX-E400 edge AI system, powered by Intel® Core™ Ultra Series 3 processors. This system adopts a CPU-GPU-NPU heterogeneous computing architecture, supporting up to 16 simultaneous video streams and parallel AI model inference. A live demonstration using Intel OpenVINO™ will showcase how this tri-architecture flexibly allocates workloads, significantly optimizing edge AI computing performance.
Meanwhile, through deep collaboration with Qualcomm Technologies, Inc., Innodisk continues to expand its AI on Qualcomm Dragonwing™ series of edge AI platforms, achieving high-performance, low-power edge inference in a compact form factor. The exhibition will feature the latest IQ9 and IQ10 high-compute AI platforms, as well as the IQ-X series edge AI solutions supporting Windows systems. Additionally, live demonstrations such as “AI Thread Defect Detection” and “Multi-Channel Visual Image Processing” are powered by Innodisk’s COM-HPC Mini platform, which uses the Dragonwing IQ-9075 SoC. Furthermore, Innodisk integrates AMRs with GMSL2 cameras and 3D depth cameras to achieve precise spatial awareness, ROI security detection, and automatic emergency braking systems for efficient real-time decision-making.
Innodisk: Hardware as a Solid Foundation, Software for Application Value-Add
Edge AI applications require a reliable, efficient, and highly dependable hardware architecture. With two decades of deep expertise in the industrial control sector, Innodisk continues to drive product innovation and lead industry standards. This exhibition comprehensively showcases its industrial-grade memory, storage, communication, and sensing product lines, further solidifying its industry leadership. In memory, it will feature next-generation DDR5 8000 RDIMM, CUDIMM, CSODIMM, as well as industry-leading 12800 MRDIMM and CXL add-in cards (AIC). Storage products include data center SSDs with EDSFF and U.2 interfaces, along with industrial-grade solutions using the latest 218-layer 3D TLC technology. In communication and sensing, Innodisk debuts the SFP28 25GbE Ethernet expansion module and the world’s first M.2 interface SFP+ network expansion module. Additionally, for AI visual recognition, the exhibition will showcase a full range of industrial-grade camera solutions with GMSL2, MIPI, and USB interfaces.
Software enablement is also a core part of Innodisk’s AI ecosystem. Innodisk has successfully integrated the end-to-end edge AI MLOps platform Edge Impulse with its self-developed cloud-based intelligent management tool iCAP, demonstrating Agentic AI capabilities for automated remote model updates and rapid deployment. Through this comprehensive hardware and software showcase, Innodisk fully presents its “Edge AI Ecosystem: Five Key Layers of Architecture,” helping customers advance their AI transformation safely, reliably, and at scale across diverse application scenarios.
We cordially invite you to visit COMPUTEX 2026 to experience Innodisk’s range of leading-edge AI solutions.
- Date: June 2 to June 5, 2026
- Location: Taipei Nangang Exhibition Center, Hall 1, Booth K0106 (Innodisk)