Samsung SDS announced on the 2nd that it has begun construction of the Gumi AI Data Center, an artificial intelligence-dedicated facility in Gumi, North Gyeongsang Province, moving into the full-scale building phase of the project.

Site, Capacity, and Timeline

The Gumi AI Data Center will be built on a site of roughly 14,000 pyeong in Gongdan-dong, Gumi, with an initial power capacity of 60 megawatts (MW). The facility has been designed so that it can be expanded in stages to a maximum of 120 MW in line with rapidly growing demand for AI.

According to the company, Samsung SDS received approval to begin construction from Gumi City at the end of August. Work is currently under way, and the company is aiming for completion at the end of 2028.

Part of Samsung's USD 44.74 Billion Yeongnam Investment Plan

The project is part of the USD 44.74 billion investment plan for the Yeongnam region that Samsung announced in July. At that time, Samsung said it would expand investment centered on several projects in order to build the Yeongnam region into a global physical AI innovation cluster.

Those projects include a humanoid robot and AI data center in Gumi, all-solid-state batteries in Ulsan, package substrates for AI servers and multilayer ceramic capacitors (MLCCs) in Busan, and high-value-added ships in Geoje.

Design Built on Three Decades of Experience

The Gumi facility was designed on the basis of Samsung SDS's 30 years of accumulated experience in building and operating data centers. It is described as a hyperscale data center optimized for running high-power IT equipment such as high-performance graphics processing unit (GPU) servers. The company plans to secure scalability and stability so that the facility can respond to growing AI demand.

To that end, Samsung SDS will build out large-capacity power infrastructure, redundancy for major equipment, and an emergency power system.

Hybrid Cooling for High-Density GPU Servers

To effectively control the heat generated by high-performance GPU servers, the data center will apply a hybrid cooling technology that combines air cooling and liquid cooling.

Air cooling will be used for general servers, such as network and storage equipment, that generate relatively little heat. Liquid cooling, which circulates water to remove heat, will be applied to high-density GPU servers.

Server rooms will be designed to accommodate racks of various power densities, and the company plans to build an environment able to stably operate ultra-high-power racks in the future. The facility is also designed so that it is not limited to specific GPUs or services and can flexibly accommodate a variety of AI systems.

Business Model: GPU Services and Colocation

Samsung SDS plans to pursue not only services in which it builds and operates its own GPUs, but also a colocation business in which customers install and use their own GPUs. Through this, the company intends to provide AI services to a range of external customers while also supporting the core businesses of Samsung affiliates.

In particular, the company plans to link the facility with its manufacturing robot transformation (RX) business and use it in fields requiring large-scale computation, such as AI training for precision-work robots on manufacturing sites.

Company Statement

Samsung SDS CEO Lee Jun-hee said, "We will gather the data center construction and operation experience we have accumulated, along with our cloud and AI technology capabilities, to create AI infrastructure that customers can trust and use."

He added, "We will contribute to the development of the regional economy and industrial ecosystem and lead the AI transformation (AX) of manufacturing and industrial sites based on next-generation AI infrastructure."

Staying Ahead of South Korea's Data Center Boom

Every hyperscale facility, colocation campus, and edge deployment across South Korea begins with an announcement, and the firms that track those announcements earliest consistently win the best contracts. The question is whether your pipeline relies on news scraps or a structured intelligence system.

The Global Project Tracking (GPT) platform by Blackridge Research consolidates data center development activity across the country into a single, searchable dashboard, giving you visibility into opportunities months before competitors even hear about them.

From land acquisition to commissioning, every stage of a project's lifecycle is captured, updated, and delivered to your team in real time.

• Upcoming Projects

• Tender Notices

• Contract Awards

• Projects Under Construction

• Completed Projects

Book a free demo today and see how the Global Project Tracking (GPT) platform by Blackridge Research can transform your approach to the South Korea data center market.