KT Cloud commits to supply 1 GW of AI data center capacity in South Korea by 2031 using modular construction, accelerating Asia-Pacific build-out.

KT Cloud plans to deliver over 1 gigawatt (GW) of artificial intelligence data center (AIDC) infrastructure across approximately 20 sites in South Korea by 2031. This expansion will add 800 megawatts (MW) to its existing portfolio of roughly 200 MW. To accelerate deployment, the company will leverage standardized modular designs and concurrent construction methods. Simultaneously, it aims to attract global AI demand to South Korea through a regional cluster strategy that spans the Seoul metropolitan area and surrounding provinces.

The roadmap was announced at the "KT Cloud Summit 2026," held on the 15th at the Grand InterContinental Seoul Parnas in Gangnam-gu, Seoul. Approximately 1,500 attendees, including public and corporate representatives alongside data center, cloud, and AI specialists, gathered to review the company’s infrastructure strategy. Presented under the theme "AX for Your Business," the initiative aims to facilitate tangible AI transformation execution.

During his welcome address, KT Cloud CEO Kim Bong-kyun outlined the dual-region approach: "In the Seoul metropolitan area, we will aggressively secure attractive sites with power and infrastructure in place, while in non-metropolitan regions we will expand supply by first securing tenant customers and then acquiring sites." He further emphasized, "By combining our accumulated cloud and AIDC technology and operational experience with KT Group's core infrastructure and our partners' specialized expertise, we will support customers so that their AI transformation execution translates into tangible business results."

Choi Ok-jin, head of KT Cloud’s business division, detailed the supply strategy, which focuses on expanding customer-specific clusters within the Seoul metropolitan area while developing large-scale AIDCs tied to global demand in non-metropolitan regions. Metropolitan sites will be segmented by client profile: the Bucheon-Cheongna cluster will cater to domestic and international big tech firms requiring large-scale AI capacity; the Mokdong cluster will target domestic and international cloud service providers (CSPs); the Yeouido cluster will specialize in ultra-low-latency financial services; and the Ansan-Yongin cluster will host an "AI GPU Zone" optimized for high-density GPU deployments.

For non-metropolitan regions, KT Cloud will follow a phased rollout, beginning with 40 MW initial projects before scaling to hundreds of megawatts. "We will actively attract global AIDC demand into South Korea," Choi stated. The delivery timeline calls for the Bucheon AIDC in 2027, followed by the Daegu, Gunsan, and Yongin AIDCs in 2028, and the Ansan AIDC in 2029.

To secure the additional 800 MW, KT Cloud has identified a 1.2 GW site pipeline. Aligning with industry shifts toward higher power density, the facilities will incorporate liquid cooling, ultra-high-density power systems, and AI-driven intelligent operations. The operational strategy also leverages digital twins and robotics for rapid failure and disaster response, while adherence to Open Compute Project (OCP) standards will ensure broad compatibility across equipment and technology vendors.

Modular AIDC construction serves as the primary catalyst for accelerated deployment. KT Cloud is collaborating with SGC to merge power and energy infrastructure capabilities with comprehensive data center design, construction, and operational expertise. This partnership aims to compress construction schedules through standardized designs and parallel workflow execution, enabling supply to scale directly with demand. As part of this initiative, SGC is investing 872 billion won (approximately $641.5 million) in the Gunsan AIDC project. The developer will construct a 60 MW first-phase facility on a 115,000-square-meter site, with construction commencing in October and commercial operations targeted for the first quarter of 2028. The project is ultimately designed to reach 300 MW, with SGC covering 100% of the capital expenditure.

Physical AI will also be integrated into AIDC operations. During a first-phase proof of concept at its AI Innovation Center, KT Cloud deployed autonomous mobile robots (AMRs) for facility inspections but encountered limitations: the units struggled with obstacle navigation and failed to seamlessly integrate collected data with operational systems. For the second-phase proof of concept this year, KT Cloud will partner with Hyundai AutoEver to deploy Boston Dynamics’ quadruped robot, "Spot," instead. "Through group-level collaboration, we plan to build a data center-specific data store and advance physical AI models," Choi noted. "Ultimately, what we want to create is a fully autonomous AIDC."

Regionally distributed AIDCs will be interconnected via KT’s ultra-low-latency backbone network. Jeon Myung-joon, head of KT’s Enterprise Service division, stressed the critical role of connectivity: "An AIDC is not complete with just data centers, computing, and data. Without connectivity linking them, a data center is merely a collection of computing resources." He added, "Infrastructure that connects South Korea's AIDCs at ultra-low latency must operate as a single ecosystem for AX to become a reality."

Within its cloud offerings, KT Cloud introduced a "Composite AI" strategy designed to unify disparate cloud environments and AI resources into a single operational framework. Gong Yong-jun, head of KT Cloud’s platform division, explained, "The reality is becoming increasingly clear that a single vendor's cloud alone can no longer solve all the problems an enterprise faces. Companies must be able to select and combine resources suited to each requirement—security, performance, cost, and data."

KT Cloud also announced it will leverage its proprietary "KT Cloud Platform" as a foundational growth engine, extending publicly validated technologies and operational expertise into the commercial market. The "Single Cloud Experience," which unifies portals, access permissions, and governance policies across multi-cloud environments, is scheduled for completion by early 2027. Following this milestone, the platform will integrate compute resources—including GPUs and NPUs—alongside AI models and task-executing AI agents into a unified execution framework. The platform’s capabilities will subsequently expand to encompass disaster recovery (DR) and advanced AI utilization.

Public-sector AX implementation case studies from North Gyeongsang Province were also highlighted. The provincial government plans to migrate 65.5% of its information systems to the cloud and launch a public-private partnership, "Gyeongbuk AI·CDC," to serve as a regional AX hub. Leveraging this infrastructure, AI applications will be deployed across four core public services: policy decision-making, administrative operations, disaster response, and care services.

The summit concluded with 20 afternoon sessions divided into two tracks: "AIDC & AX Infrastructure" and "Cloud Platform & Services." An accompanying exhibition featured emerging technologies and services across three categories: AIDC, the KT Cloud Platform, and AI APIs. Industry partners, including Rebellions, Rableup, Didim, Dell Technologies·Etech System, and EPI Korea, also showcased AX-focused technologies and solutions.