The useful question is what changes for users, developers or buyers, and whether the announcement stays industry context or becomes something people can actually use.

AI data centers differ from traditional facilities by processing vast datasets in real time to generate new outputs, requiring specialized hardware like GPUs, NPUs, and HBM memory. These systems operate at 5–10 times the power and heat output of conventional data centers, necessitating advanced cooling and power management. SK Telecom’s initiative addresses these challenges by integrating high-performance computing with scalable infrastructure to support AI workloads efficiently.

The company will establish regional AI infrastructure hubs starting in Ulsan, expanding to Yeongnam, Chungcheong, and Honam by 2035. SK Telecom plans to open 5GW of capacity annually from 2029, scaling to 15GW total. This expansion aims to decentralize data center locations beyond the Seoul metropolitan area, fostering regional development and reducing infrastructure bottlenecks.

SK Telecom created a dedicated entity, SKhypre, to oversee the project, with a 750 billion won investment by 2030. SKhypre will prioritize securing shared infrastructure such as substations, power transmission lines, and grid connections to accelerate deployment. The company positions AI data centers as critical national infrastructure, comparable to high-speed internet and highways, to bolster South Korea’s AI ecosystem.

Global demand for AI infrastructure is rising, with South Korea’s stable power supply, semiconductor supply chain, and industrial expertise cited as competitive advantages. The International Energy Agency projects global data center power consumption to surpass that of major developed nations within years, underscoring the urgency of efficient energy solutions in AI data center operations.