SK Group seeks 464 trillion won for massive AI data center expansion

A massive infrastructure push

SK Group is currently in discussions to attract significant investment from global big tech companies and financial partners. The goal is to construct a massive network of AI data centers to meet the growing market demand for computing power. According to 머니투데이, the company is looking to establish a significant footprint in the Korean market first.

The scale of this ambition is substantial, with a plan to build a total of 15GW of data center capacity by 2035. To reach this goal, the company will first focus on a 5GW phase. This initial phase is estimated to require approximately 464 trillion won, which covers the necessary equipment and construction costs.

SK Hyper, a subsidiary of SK Telecom established in July this year, will lead these efforts. The company intends to create a data center cluster of more than 2GW in the Yeongnam region. This will begin with the Ulsan Artificial Intelligence Data Center (AIDC) and expand to other regions like Chungcheong and the Southwest.

The cost-sharing model

Because the capital required for these projects is so high, SK Group is not planning to bear the entire cost alone. Instead, the company is developing a structure based on cost-sharing between external investors and the companies that will eventually move into the data centers. This approach helps distribute the financial risk across multiple stakeholders.

SK Hyper expects to secure its first round of external funding by the end of this year or early next year. Following that initial funding, the company plans to conduct additional fundraising efforts over the next three to four years. This phased approach allows the group to manage the massive capital expenditure required for the infrastructure.

The company is also being cautious due to growing theories regarding an AI bubble. To manage risks, SK Group has decided not to deploy the entire investment amount at once. They plan to adjust the pace of their investment based on actual market demand to ensure they can secure profits that match their spending.

Overcoming high energy costs

One of the primary challenges for building data centers in South Korea is the relatively high cost of domestic electricity. To address this, SK Group plans to use a strategy that combines the technologies of its various affiliates. By integrating expertise in semiconductors, energy, and telecommunications, they aim to improve computing efficiency.

Improving efficiency is a key way to lower the operational costs of the data centers. By making the hardware and software work together more effectively, the company can reduce the amount of power needed for each unit of computing. This technical synergy is a core part of their plan to remain competitive in a high-cost environment.

The company is also consulting with the government to secure additional sites for these facilities. This collaboration with the public sector is intended to ensure that the infrastructure has the necessary space and support to grow. The goal is to create a nationwide AI infrastructure that can support both local and international needs.

A regional AI hub

SK Group is not just looking at the domestic market but is also targeting demand in Asian markets like Japan and Taiwan. The company wants to induce global tech companies to use South Korea as a primary hub for AI training and service operations. This would position the country as a central node in the global AI supply chain.

Beyond just providing space, the company is considering a business model that provides integrated AI computing resources and AI models to governments. This could include providing services to regions in Asia and the Middle East. By offering a full suite of services, they hope to become a primary partner for public sector AI initiatives.

While the target for completing the initial data centers is 2029, the company expects to begin generating revenue as early as next year. This suggests that the business model will involve early-stage services or pre-sales. The ultimate goal is to create a scalable operation that can expand to 15GW as global customer demand continues to rise.