SK Telecom will spin off its subsidiary SK Broadband's artificial intelligence (AI) data center business into a separate corporation. The move is seen as an effort to consolidate AI data center-related organizations and assets scattered across the two companies to speed up fundraising and business expansion.
According to the industry on the 27th, SK Telecom and SK Broadband each held a board meeting that day and approved a proposal to spin off the AI data center business unit. The two companies plan to finalize the specific spin-off method, the new entity's equity structure, and the launch timeline, and then proceed with the related procedures.
The analysis is that the spin-off envisions bundling SK Telecom's AI data center-related business and SK Broadband's data center business into a separate corporation. Details such as assets and personnel subject to transfer before the spin-off are also expected to be finalized in stages.
SK Telecom and SK Broadband have each operated AI data center organizations. SK Telecom handled AI cloud and graphics processing unit (GPU) services and data center business development, while SK Broadband was in charge of building and operating data centers nationwide and customer sales. However, there has been criticism that, because the entities are split into a parent and a subsidiary, there are limits to centralizing large-scale investment and business decision-making.
Last month, SK Telecom launched the AI DC integration task force reporting directly to the chief executive officer (CEO), kicking off a full-fledged reorganization of related units. It then established SK Hyper, a separate corporation responsible for securing AI data center sites and power and attracting global customers. The spin-off is seen as a step to reorganize the development, investment, and build-and-operate functions of the AI data center business into an independent business system.
The industry views the spin-off as a preliminary step to attract large-scale external capital. AI data centers require massive funding, from securing sites and power to purchasing GPUs and building cooling facilities. Separating the business into a standalone corporation, rather than keeping it within a telecommunications company's internal organization, makes it relatively easier to bring in strategic or financial investors and to use project financing.
AI data centers are the fastest-growing area within SK Telecom's business. Second-quarter AI data center revenue this year was 136.2 billion won, up 92.5% from the same period a year earlier. SK Telecom plans to gradually operate AI data centers totaling 5 gigawatts (GW) at major hubs nationwide starting in 2029 and expand that to 15 GW by 2035.
SK Telecom is also working with Nvidia to build the next-generation AI data center, the AI Factory. An AI Factory is a facility that combines GPUs, high-speed networks, and power and cooling systems to produce the computing resources needed for AI training and inference. SK Telecom plans to start operating the AI Factory in 2027 and expand it in stages to the GW level. The spin-off is seen as a move to clarify the execution entity that will exclusively handle investment, construction, and operation of such mega-scale AI infrastructure projects.
SK Group is pursuing an AI infrastructure business that combines capabilities in semiconductors, energy, construction, and telecommunications, starting with the Ulsan AI data center. In Ulsan, it is building the first AI data center with Amazon Web Services (AWS), aiming to begin operations in the second half of 2027, with a long-term plan to expand beyond 1 GW. SK hynix will supply AI Semiconductor and high bandwidth memory (HBM), the energy affiliates will supply power, and SK Telecom and SK Broadband will handle data center construction and operation.
An industry official said, "AI data centers differ from existing telecom businesses in investment size, business cycles, and financing methods," and added, "If it becomes independent as a separate corporation, it will likely speed up decision-making when pursuing business with global big tech and investors."