Hyundai E&C to complete S. Korea's first AI-dedicated data center in 16 months

Hyundai Engineering & Construction said Monday it has finalized a main contract for the construction of the Pohang AI data center — South Korea's first AI-dedicated data center — and begun installing precast concrete components, marking the official start of the project.

The project will build a GPU-based, high-performance data center with a power receiving capacity of 40 MW at the Gwangmyeong General Industrial Complex in Ocheon-eup, Nam-gu, Pohang, North Gyeongsang Province. Power receiving capacity refers to the maximum electricity a facility can draw from the external grid. At 40 MW, the center would draw enough power for more than 130,000 households simultaneously, based on the average monthly electricity consumption per household in Pohang's Nam-gu of 214.78 kilowatt-hours.

Unlike conventional data centers built around general-purpose computing, the facility is designed specifically for AI workloads — real-time analysis and processing of massive datasets — using high-performance GPU infrastructure. Construction began in June, with operations targeted for the second half of 2027. The total construction period is 16.5 months.

To meet the tight schedule, Hyundai E&C applied precast concrete construction across the entire project. While some domestic commercial data centers have used precast methods for select structural elements such as beams and columns, this marks the first time the technique has been extended to all concrete components — including slabs, walls and staircases.

The approach cuts framing work that would normally take about a year under conventional cast-in-place methods down to two months. Hyundai E&C said it also expects the method to improve safety and quality alongside the shortened timeline.

The company also applied a pre-construction service — reviewing the optimal project structure and construction methods before breaking ground — alongside a fast-track approach that ran design, permitting and construction preparation in parallel. That combination allowed Hyundai E&C to break ground within a month of joining the project.

Energy efficiency was also a priority. The facility will be the first in South Korea to adopt a water-cooled cooling system designed to handle the high heat output of AI servers. Its power usage effectiveness, an industry benchmark for energy efficiency, was set at 1.25 to minimize energy loss.

The project developer, Neo AI Cloud, plans to build the Pohang AI data center as a scalable infrastructure capable of accommodating next-generation GPUs. The company intends to link the facility with GPU-as-a-service and AI platform services to support AI model development and service operations for domestic and international clients. GPU-as-a-service allows companies to rent high-performance GPU hardware through the cloud on demand, without purchasing the equipment outright.

Hyundai E&C said it aims to strengthen its construction competitiveness in the AI data center market through this project. "With the spread of generative AI and large-scale AI models, competitiveness in the data center market is shifting beyond scale toward what the industry calls 'time to market' — the race to launch services faster," a company official said. "We will successfully complete the Pohang AI data center on schedule and demonstrate differentiated competitiveness that secures business viability, construction timelines and quality all at once, alongside our best-in-class construction capabilities."

Hyundai E&C has built 19 data centers in South Korea — more than any other domestic construction company — starting with the Korea Financial Telecommunications & Clearings Institute's Bundang center in 2004, followed by KT Corp's Mokdong IDC, the Yongin Pacific Sunny data center and Naver's data center "Gak Sejong." The company is also the only Korean builder to have constructed data centers for global big-tech firms, giving it globally competitive technical expertise.

Going forward, Hyundai E&C plans to develop data center models integrating renewable energy and small modular reactors, tailored to the energy characteristics of each region, by leveraging its strengths in the energy sector. The company aims to provide a total solution spanning the entire energy value chain, advancing core AI-era infrastructure and driving energy innovation at the same time.