SK Telecom (SKT) plans to build a 15GW AI data center portfolio in South Korea, with 5GW coming online in phases starting from 2029. The scale of the project - whose total cost for a single 1GW facility could approach KRW 70 trillion - aims to position the country as Asia's primary AI infrastructure hub amid a global shortage of data center capacity.
The company will expand its under-construction Ulsan site into a cluster exceeding 2GW across the southeastern Gyeongsang region, then add another 1GW in the southwestern Jeolla region. The combined 5GW domestic footprint forms the first stage of a longer-term 15GW buildout. SKT said financing would draw on its own investment, strategic partners, long-term customer contracts, and project financing.
A supply gap drives the investment
Global demand for data center capacity is forecast to grow 19-22% annually, with supply failing to keep pace, according to McKinsey & Company. The consultancy estimates a shortfall of roughly 15GW in the U.S. alone by 2030. That gap is pushing big tech firms to look beyond North America for AI infrastructure sites.
South Korea has several attractions: it is a leading producer of high-bandwidth memory (HBM), operates stable power grids built on nuclear and LNG generation, and possesses GW-scale infrastructure expertise honed through decades of running semiconductor fabs. SKT expects these advantages to draw hyperscaler tenants into the new facilities.
Full-stack delivery through SK Group affiliates
SK Group's affiliates already cover the three pillars of AI data center infrastructure - semiconductors, energy solutions, and construction/operations. SKT will act as the lead architect, coordinating design, build, and operation across the program. The company previously outlined an "AI Factory" model, a next-generation data center concept that it aims to start operating in 2027 before scaling to gigawatt proportions.
"This AI data center project is aimed at preemptively preparing the computing infrastructure that the global AI ecosystem needs," said Jung Jai-hun, President and CEO of SKT. "We will work closely with the government, industry, and local communities to help Korea grow into Asia's core AI infrastructure hub."
Tied to national infrastructure and regional development
SKT is framing the data center push as Korea's third national infrastructure revolution, following the Gyeongbu Expressway (1968) and high-speed internet (1998). The company is aligning the buildout with the government's "AI G3" strategy, which targets a top-three global AI power ranking, and with policies for balanced regional development.
The initiative links data center siting with local industry, power availability, and anchor tenant commitments. It also dovetails with Seoul's strategic supply plans and efforts to channel AI demand into domestic infrastructure rather than overseas capacity.
Why this matters for IT and development professionals
Data center capacity constraints directly affect cloud service availability, GPU instance provisioning, and training pipeline timelines. The 15GW roadmap signals a substantial future supply of compute in Asia-Pacific, which could shift workload placement decisions and latency profiles for teams running inference or training at scale. For engineers building distributed systems, understanding where next-generation capacity will emerge helps anticipate operational design choices - from data residency to failover planning. As AI infrastructure evolves into a core IT and development skillset, the physical buildout behind it becomes more than a facilities story; it's a factor in architecture roadmaps.
Your membership also unlocks:
