Key Highlights

1. A Bold Multi-Year Buildout Plan

SK Telecom (NYSE: SKM) aims to scale AI data center capacity from an initial 5GW in 2029 to 15GW by 2035, one of the more ambitious infrastructure roadmaps announced by a telecom operator.

2. Execution Risk Looms Large

The long timeline and significant capital commitment, including a ₩397 billion investment in chip affiliate SKHNPS, raise questions about whether a traditional carrier can compete against dedicated hyperscale operators.

3. Market Response Remains Cautious

Shares edged only modestly higher in after-hours trading, reflecting investor uncertainty about the plan's near-term financial impact given its multi-year horizon.

SK Telecom's (NYSE: SKM) newly unveiled plan to build up to 15 gigawatts of AI data center capacity by 2035 raises a fundamental question for investors: can a telecom carrier, built primarily around wireless and fixed-line network operations, successfully compete against dedicated hyperscale operators that have spent years refining their data center development and operational expertise?

The scale of the ambition is significant. Starting with 5GW of capacity from 2029 and tripling to 15GW by 2035, the roadmap would place SK Telecom among the more consequential AI infrastructure providers globally, assuming successful execution. The company is also backing the strategy with a ₩397 billion capital injection into SKHNPS, its AI chip affiliate, signaling a coordinated push across both compute infrastructure and the underlying semiconductor supply chain.

However, building and operating data centers at hyperscale requires a different set of competencies than running telecom networks, including expertise in cooling systems, power procurement at massive scale, and enterprise-grade service level agreements with AI and cloud customers. Dedicated hyperscalers with years of experience in this domain represent formidable competition, and SK Telecom will need to demonstrate it can attract anchor customers willing to commit to long-term capacity agreements well before the first phase comes online in 2029.

The multi-year timeline itself introduces execution risk. Significant capital expenditure programs stretched over a decade are vulnerable to shifts in technology, changes in AI compute architecture, and potential cost overruns, all of which could affect the ultimate return on investment. Investors have so far responded cautiously, with shares edging only modestly higher in after-hours trading following the announcement, reflecting a wait-and-see posture.

For the plan to be viewed as a success by the market, SK Telecom will likely need to provide more granular details over time, including financing structure, expected returns on invested capital, and early signs of customer demand, before investors fully price in the AI infrastructure opportunity into the stock's valuation.