Elice Inc. Joins Korea's First Industrial-Complex Edge AI Data Center Powered by GPU & NPU

(Note: The image above is an AI-generated illustration to aid understanding of the modular AI data center ‘Elice AI PMDC’ and may differ from the actual deployment site.)

Supplies modular AI data center (AI PMDC) and AI infrastructure virtualization solution (ECI) as a package to Busan Shipbuilding & Marine Industrial Complex

AI full-stack company Elice Inc. (CEO Jae Won Kim), providing AI infrastructure, cloud, and industry-specific solutions, has been selected to participate in the “Busan Shipbuilding & Marine Edge AI Data Center Establishment Project” led by the Korea Industrial Complex Corporation (KICOX) Busan Branch. This project was chosen for the Ministry of Trade, Industry and Energy’s “2026 Industrial Complex Edge AIDC Demonstration Pilot Project.” It marks the nation’s first industrial-complex edge AI data center built on GPU and domestic NPU, aimed at accelerating AI transformation (AX) for manufacturers.

Through this project, Elice Inc. secures its first commercial reference supplying the modular AI data center “Elice AI PMDC” and the AI cloud infrastructure virtualization solution “ECI” as an integrated package to an external customer—both validated on its own Elice Cloud. This demonstrates Elice Inc.’s technical capability as an AI infrastructure company spanning both hardware and software.

Elice Inc. will deploy two units of AI PMDC, which embody high-density power design technology, and provide the ECI integrated operation platform for efficient control. ECI’s technology for allocating and operating heterogeneous AI accelerators—GPUs and domestic NPUs—together will maximize compute utilization and deliver a low-latency, high-efficiency AI environment optimized for manufacturing sites.

Accelerates manufacturing AX through integrated operation of domestic NPU/GPU and DR-linked architecture

The center will operate as Korea’s first industrial-complex edge AI data center model combining GPU and domestic NPU: GPUs handle training, validation, and high-performance computing for manufacturing AI models, while domestic NPUs manage inference and SaaS services. By leading the establishment and operation of NPU-based AI infrastructure, Elice Inc. aims to support on-site deployment of manufacturing AI services and contribute to expanding the domestic AI semiconductor ecosystem.

For resilience, Elice Inc. will link the edge AI data center with Elice Cloud in a disaster recovery (DR) architecture. A redundant, uninterruptible power supply and dual-layer security system will ensure continuous AI service availability even during failures.

Elice Inc. CEO Jae Won Kim emphasized

“The key lies in the organic combination of custom data center technology that accommodates high-performance AI infrastructure hardware and the operating software that controls it efficiently. This project marks a major milestone as the first application of the AI PMDC and ECI integrated package—validated on our own AI cloud—to an industrial site, serving as a critical stepping stone for both manufacturing AI transformation and the proliferation of the domestic AI semiconductor ecosystem.”