Earnings call transcript: SK Telecom’s Q1 2026 results show subscriber gains

SK Telecom reported Q1 2026 consolidated revenue of KRW 4.39 trillion, up 1.5% quarter-over-quarter, and operating income of KRW 537.6 billion as wireless subscriber net additions rebounded by about 210,000. AI data center revenue rose 89% year over year, and the company resumed dividends at KRW 830-831 per share. Management guided for further earnings recovery in 2026, though the stock fell 2.43% in pre-market trading.

Analysis

The earnings reset changes the competitive frame more than the headline numbers suggest. A return to positive subscriber adds plus a resumed cash payout implies the company is trying to re-establish a “quality incumbent” regime, which usually hurts smaller Korean mobile challengers more than it helps the incumbent’s own top line; in saturated telecom, share gains are often zero-sum and the real swing factor is churn economics, not gross additions. The more important second-order effect is that management is explicitly signaling a willingness to defend high-LTV customers rather than buy growth at any price, which should support margin durability if the industry rationalizes promotions.

The AI/data-center angle is the real option embedded in the story. If the AI data center business is already contributing meaningfully to growth while management frames profitability as at least comparable to legacy telecom, the market may be underestimating the mix shift from low-growth regulated cash flow to a structurally tighter capacity market with pricing power. That said, the absence of disclosed margin metrics is a tell: the current narrative is still mostly about revenue visibility and strategic positioning, so the stock can stay range-bound until investors see evidence that incremental utilization is translating into operating leverage rather than capex drag.

Near-term downside is less about the quarter and more about expectations reset. After a strong six-month rerating, even “good” results can sell off if the path to a larger dividend and AI monetization is deferred by one or two quarters. The key catalyst sequence is: stabilize subscribers over the next 1-2 quarters, show AIDC capacity ramp without margin leakage into year-end, then use the 2026 year-end dividend as proof that earnings normalization is durable.

The contrarian view is that the market may be overly fixated on the payout resumption and missing that this is still a transition story, not a clean ex-cyber incident re-rating. If marketing intensity rises to defend share, or if AI infrastructure capex outpaces revenue conversion, the equity could trade like a value trap with a growth veneer. On the other hand, if management keeps discipline and the AIDC buildout scales as implied, the current multiple leaves room for upside because investors have not yet been given the metrics that would justify paying up for the AI segment.

Key Decisions for Investors

• Long SKM on pullbacks over the next 2-6 weeks, but size modestly; use the post-earnings weakness as an entry if the stock holds above the recent pre-market low. Risk/reward is attractive if the market is over-discounting the path to dividend normalization and AI monetization.
• Pair trade: long SKM / short a Korean wireless competitor or regional telecom proxy for 1-3 months. Thesis: SKM is better positioned to convert customer recovery into dividend credibility, while incumbents with weaker AI optionality should lag if the industry becomes less promotional.
• Sell downside protection on SKM via put spreads into the next 1-2 quarters if implied volatility stays elevated. The catalyst stack is gradual rather than binary, so premium decay should favor investors who believe the subscriber recovery is sustainable.
• If you want AI exposure with less beta than semis, initiate a tactical long SKM as a data-center/AI-infrastructure proxy for 3-6 months. The trade works only if the market starts valuing contracted capacity and not just handset growth.
• Avoid chasing NVDA off this print; the article is more supportive of AI infrastructure demand than near-term GPU upside. Use NVDA only as a relative-value leg versus SKM if you want to express that infrastructure monetization is still underappreciated.