Innodisk at COMPUTEX 2025: Scalable Edge AI, Private LLMs, and Cross-Platform Readiness for Industrial AI

At COMPUTEX 2025, Innodisk reaffirmed its leadership in industrial-grade AI enablement with a strong focus on real-world deployment.

Enabling AI Across Heterogeneous Compute Platforms

Innodisk’s strength lies in its ability to deploy seamlessly across fragmented hardware platforms. At the show, the company demonstrated complete AI systems built on NVIDIA, Intel, and Qualcomm Technologies architectures—augmented by its in-house memory, flash storage, camera modules, and expansion peripherals.

This highlights Innodisk’s value proposition as a platform-agnostic AI infrastructure enabler. Whether the application is vision-based analytics in transportation or multimodal recognition in smart cities, Innodisk ensures system-level compatibility across diverse industrial environments.

On-Premise LLM Training and Inference: APEX-X100 with AccelBrain

One of the key announcements was the APEX-X100, Innodisk’s short-depth AI computing platform for private LLM training and deployment.

• Powered by AccelBrain, Innodisk’s proprietary AI stack
• Designed for data-sensitive sectors such as healthcare, defense, and manufacturing
• Eliminates the need for external cloud dependency, supporting sovereign AI strategies

This edge-native system provides IT leaders with a practical pathway to retain control over proprietary models and inference pipelines—an increasingly critical consideration in enterprise AI governance.

Short-Depth AI Server: APEX-S100

For edge deployments with space constraints, Innodisk introduced the APEX-S100, a 2U 420mm-deep server supporting:

• Two double-width GPUs
• Advanced thermal control
• Flexible DRAM and NVMe expandability

This system is ideal for AI inference at industrial edge sites, such as logistics hubs, manufacturing floors, and telco nodes.

AI Vision Systems with Modular Camera Architecture

Innodisk’s updated embedded vision lineup now includes camera modules supporting:

• USB, MIPI, and GMSL2
• Custom adapter boards and drivers across Raspberry Pi, industrial PCs, and AI servers

A joint demonstration with Advantech’s Rugged Edge AI Division showed real-time processing across 16 GMSL2 camera channels, integrated into TREK-50N and REX-52N systems. The target use case: surround-view monitoring for industrial vehicles and driver behavior tracking—vital for workplace safety in mining, logistics, and fleet operations.

Qualcomm Collaboration: Edge-Ready Smart Parking AI

Through its ongoing collaboration with Qualcomm Technologies, Innodisk also showcased a smart parking and security solution leveraging:

• Qualcomm Dragonwing™ IQ9
• Built-in NPU for low-latency multimodal processing
• Real-time inference for vehicle make/model/color recognition and behavioral analytics

With fully local inference, camera expansion support, and VLM/GenAI compatibility, this solution is designed for smart city deployments, where security, scalability, and responsiveness are mission-critical.

Reference Kits for Intel® Core™ Ultra Series 2

Innodisk’s reference kit for Intel Ultra Series 2 provides an out-of-the-box solution for OEMs and system integrators:

• Includes MIPI-over-Type C camera modules, LPDDR5X DRAM, Gen5 SSDs
• Fully modular with expansion options
• Supports AI, automation, and data visualization tasks across manufacturing and public sector projects

AI-Ready Memory & Storage Innovations

As AI models scale, so too must the memory and storage behind them. Innodisk’s updated lineup includes:

• LPDDR5X CAMM2 and DDR5 MRDIMM modules for high-density edge and rack systems
• PCIe Gen5 SSDs across form factors: U.2, E1.S, E3.S, E3.L
• OCP SSD 2.0 compliance, ensuring interoperability and scalability for hyperscale and telco deployments

These are infrastructure-grade solutions aimed at next-generation data centers, inference clusters, and autonomous industrial systems.

Strategic Takeaway for CTOs and Infrastructure Planners

Innodisk’s 2025 showcase reflects a deliberate shift from component provider to AI systems enabler. Its continued investment in platform compatibility, modularity, and sovereign compute aligns strongly with modern infrastructure priorities:

• Edge-native AI with enterprise-grade hardware reliability
• Vertical-specific use cases (e.g., transportation, smart cities, automation)
• Cloud-optional design for compliance and resilience