SK Telecom has launched an aggressive plan to build AI data centers with a combined capacity of 15 gigawatts by 2035, aiming to transform South Korea into a major AI computing hub in Asia. The project’s scale is massive, comparable to the power consumption of some small countries, signaling SK Telecom’s intent to become a foundational player in AI infrastructure.
Partnership with Nvidia and Initial Investments
The telecom giant’s parent, SK Group, joined forces with Nvidia in early June 2026 to tap Nvidia’s DSX platform for developing AI cloud infrastructure at gigawatt scale. SK Telecom created a new subsidiary, SK Hyper, in July 2026 to steer this initiative, securing board approval for an initial investment of about KRW 750 billion (roughly $508 million), targeted for deployment by 2030.
The first phase aims for 5 GW of capacity starting operations in 2029, based in Ulsan, before expanding into the Chungcheong and Honam regions. Building a single 1 GW facility is estimated to cost approximately KRW 70 trillion, underscoring the enormous capital demands behind the full project.
This push is more than just corporate ambition. SK Telecom positions the expansion as vital national infrastructure, building on South Korea’s existing semiconductor strengths from Samsung and SK Hynix to create a tightly integrated AI supply chain.
While the initial $508 million investment marks an important commitment, it represents only a fraction of the total funding needed. Future financing will likely involve strategic partners and client contracts, as outlined in SK Telecom’s June 30, 2026 6-K regulatory filing.
Execution hurdles remain sizeable: scaling power generation, upgrading grid capacity, navigating regulatory approvals, and managing construction timelines all pose risks. Yet, if successful, this infrastructure could redefine Asia’s data center landscape and bolster South Korea’s role in global AI development.
