[E-Daily Reporter Yun Junghoon ] South Korea’s three major telecom companies are undergoing a strategic transformation to become “comprehensive AI infrastructure providers,” moving beyond their core mobile communications business to focus on AI data centers (AIDCs). While traditional data centers were essentially businesses that leased server space and power, AIDCs combine power, networks, GPUs, and cloud services to provide AI computing resources as well. This is why telecom companies are positioning AIDCs as their next-generation growth engine.

These efforts are also yielding results in financial performance. According to the telecommunications industry on the 20th, SKTelecom’s AIDC revenue for the first half of this year reached 267.6 billion won, a 91% increase compared to the same period last year. Revenue for KT Cloud—KTCorporation’s subsidiary dedicated to data centers—rose 10% to 515.5 billion won during the same period. Revenue in LG Uplus’s enterprise infrastructure division, which includes AIDC, also increased by 29% year-over-year to 238.5 billion won.

Investment is also on the rise. SKTelecom plans to gradually expand its power capacity, centered on “SK Hyper,” a newly established subsidiary specializing in AIDC. The company highlights its strength in combining the group’s capabilities—ranging from SK hynix’s semiconductors to SK Eco Plant’s construction and energy services, and SK Broadband’s operations. KTCorporation is also securing additional power capacity for its data centers. LG Uplus is expanding its investments with its Paju AIDC as a base.

The concept that telecom companies are focusing on in AIDCs is the “token factory.” This is a pricing structure where costs are charged based on the generation and usage of tokens—the basic units of AI computation. Just as a power plant generates electricity, an AIDC produces AI computing resources and supplies them to corporate customers. Telecom companies can also leverage the billing and settlement capabilities they have built up over decades.

The network resources owned by telecom companies—such as fiber-optic cables and submarine cables—are also a competitive advantage. Moving and training large-scale data across distributed AIDCs incurs enormous communication costs. Even when transmitting data from the Seoul metropolitan area to regional AIDCs, the cost burden increases significantly without a dedicated network. Connecting to overseas AIDCs also incurs additional costs for submarine cables.

Telecom companies can combine their own networks with AIDCs to provide dedicated lines. This means they can bundle not only power and computing resources but also data transfer into a single product. This is a key point of differentiation from existing data center operators.

However, there are many hurdles to overcome. The biggest challenge is power. AIDCs must secure environmental and local government permits starting from the site selection process, and they must reliably secure massive amounts of power.

Operating a 1 GW AIDC requires power equivalent to that of a large nuclear power plant. Compounded by the concentration of power demand in the Seoul metropolitan area and delays in the construction of transmission lines, it may take a considerable amount of time before actual power supply is secured.

There is also the issue of manpower. This is why competition to secure AIDC specialists—in fields such as electricity, cooling, and facilities—is already heating up. It requires personnel not only to build the facilities but also to integrate and operate ultra-high-voltage power systems, cooling equipment, GPUs, and networks.

Lee Jeong-ho, a professor in the Department of Mechanical Engineering at Ajou University, pointed out, “Although domestic companies and the government have set a target of 18.4 GW by 2035, the reality is far from easy,” adding, “It will be difficult to achieve this goal without the necessary power grid, infrastructure, and dedicated engineers.”

An official from the telecommunications industry stated, “AIDC is not simply about leasing buildings; it is a comprehensive infrastructure business that requires power, cooling, AI computation, and ultra-high-speed data transmission networks to be seamlessly integrated to function.” The official added, “Telecom companies that overcome bottlenecks in power and manpower and internalize the entire process of AI operation will secure a leading position in the high-value-added ecosystem of the next-generation AI industry.”