Skip to content

DEUS unveils modular AI Edge Cube integrating power, cooling and IT

DEUS unveiled its all-in-one modular AI edge data center, DEUS AI EDGE CUBE, at Data Center Korea 2026, with a reference configuration accommodating 144 GPUs at a 300 kW IT design load.

Share:
DEUS unveils modular AI Edge Cube integrating power, cooling and IT
Photo: www.hvacrj.co.kr
Published:

Integrated modular data center

DEUS, a full-stack data center platform company led by CEO Ryu Gi-hun, participated in Data Center Korea 2026 and unveiled its all-in-one modular AI edge data center, DEUS AI EDGE CUBE, for the first time. The product integrates power, cooling, IT, fire protection, and monitoring and control equipment into a single module.

AI EDGE CUBE, which DEUS developed and commercialized over the past two years, brings together the technology and expertise the company accumulated through data center planning and design, equipment installation, operations and maintenance. Under the concept “One Module, One Supply Window,” it handles everything from design through operational handover under a single contract and accountability framework.

The all-in-one modular AI edge data center, DEUS AI EDGE CUBE, targets service launch within approximately six months, covering the process from contracting through on-site commissioning.
The all-in-one modular AI edge data center, DEUS AI EDGE CUBE, targets service launch within approximately six months, covering the process from contracting through on-site commissioning. · Photo: hvacrj.co.kr

The product integrates core data center equipment into a single nonstandard 40-foot-class module. After factory production, integrated factory acceptance testing, or FAT, verifies interoperability between systems before shipment. The module is designed to enter operation after on-site connections and commissioning. This reduces the burden of installing and coordinating individual systems and supports on-site AI infrastructure deployment.

Liquid cooling and GPU capacity

The cooling system uses direct liquid cooling, or DLC, to support high-density GPU environments. The design load per GPU rack is 135 kW, with a target of handling 90% of the GPU heat load through liquid cooling. Four in-row cooling units in an N+1 configuration handle the remaining heat.

The reference configuration presented by DEUS has an IT design load of 300 kW and accommodates a total of 144 GPUs using 18 B300 servers with eight GPUs each. The coolant distribution unit, or CDU, has a liquid cooling capacity of 600 kW.

Integrated monitoring and service launch target

For operations, a data center infrastructure management system, or DCIM, provides integrated monitoring and control of power, cooling, environmental, security and IT information. A digital twin visualizes heat and airflow, while an open API enables integration with customers’ infrastructure-as-a-service, or IaaS, and GPU-as-a-service, or GPUaaS, environments.

A company representative said, “AI competitiveness depends on how quickly and reliably GPU infrastructure can be put in place when it is needed. After contract signing and specification finalization, DEUS AI EDGE CUBE aims to launch service within approximately six months by carrying out factory production and site infrastructure preparation in parallel, followed by integrated testing, shipment and transportation, and completion of on-site testing and integration with monitoring and control systems.”


Original Korean article: hvacrj.co.kr

English brief by Korea Wind Intelligence, koreawindintel.com

Found this useful?

Share this article with your network.