SK Telecom Co., Ltd. (NYSE: SKM) — A Regulated Utility in an AI Costume, and the Market Bought the Costume

Independent fundamental research · Date: 2026-07-19

Issuer: SK Telecom Co., Ltd. (“SKT”, “SKM”, “the company”) Ticker: NYSE: SKM (ADR); primary listing KRX: 017670 CIK: 0001015650 (foreign private issuer — 20-F / 6-K filer; no 10-K/10-Q/DEF 14A; Section 16 exempt) Sector: Communication Services — Telecommunications Services Report date: 2026-07-19 · Coverage: INITIATION (no prior report on SKM) Price: US$30.62 per ADS (close 2026-07-17, AZI) · 52-week range $19.68–$46.00 (closing) ADR ratio: 1 ADS = 5/9 (five-ninths) of one ordinary share (par ₩100); 1 ordinary share = 1.8 ADS (FY2025 20-F cover) Ordinary shares outstanding: 212,982,275 (214,790,053 issued less 1,807,778 treasury; 20-F, 2025-12-31) Market cap: ≈ ₩17.1 trillion (≈ US$11.7B at USDKRW 1,460; ordinary parity ≈ ₩80,469) Net debt: ≈ ₩7.4 trillion (₩7,357.0B YE2025; ₩7,308.8B at Q1-2026) ≈ 1.6× corrected EBITDA Enterprise value: ≈ ₩24.5 trillion (≈ US$16.8B) Fiscal year-end: December 31 · Reporting: KRW, K-IFRS (20-F financials reconciled to IFRS as issued by the IASB) Latest primary documents: FY2025 Form 20-F filed 2026-04-29 (acc. 0001193125-26-188763); Q1-2026 preliminary-results 6-K filed 2026-05-07 (acc. 0001193125-26-210347)

Basis conventions used throughout: Consolidated figures are in KRW billions unless stated otherwise. Per-share figures state their basis — per ordinary share (₩) or per ADS (US$) — with conversion at 1 ADS = 5/9 ordinary and USDKRW 1,460 (mid of the 1,443–1,488 range pulled across sources 2026-07-16/19). ROIC.ai per-share fields are per-ADS in KRW and its FY2025 EBITDA/FCF-yield fields are broken for SKM (see Section 6); all multiples herein are rebuilt by hand from 20-F/EDGAR figures and the live price.

Disclaimer: This report is the author’s own independent opinion and general information, published for informational purposes only. It is not investment advice, not an offer or solicitation, and carries no recommendation or price target anywhere outside the clearly-labeled Claude’s Take block below.

⚡ Claude’s Take

Claude’s own subjective opinion — the author’s independent view, offered as general information, not investment advice. The analytical body (Sections 1–15) below carries no position — no BUY/SELL, no price target — and stands on the evidence alone. This block is the one labeled exception.

Verdict: HOLD — own the utility, not the pyramid. Buy the AI-euphoria discounts, sell the AI-euphoria spikes. Not a short.

At $30.62 the ADR sits at the top of its base zone ($27–31). Underneath the story is a regulated Korean wireless utility — #1 for roughly forty years inside a state-designed three-license oligopoly — that is worth roughly $16–20 per ADS on a no-growth earnings-power-value basis (₩9.0–11.3T of equity EPV from the ~₩1.7T regulated operating-profit pool). Today’s buyer therefore pays ₩5.8–8.2T (~$4.0–5.6B) — 34–48% of the entire equity value — upfront for an AIDC option whose margins management refuses to disclose while claiming they are “comparable to telecom,” whose flagship project (Ulsan, ~₩7T) is funded mostly by AWS, whose anchor-customer agreements (OpenAI “Stargate Korea”) are MoU-stage with no committed capital, and whose ₩5T-by-2030 AI revenue target requires roughly 7–10× growth into a simultaneous build by Naver, KT, LG, Samsung SDS, the Korean state, and the hyperscalers. At the December 2025 low (~$19.68) that option cost essentially zero; at the June 2026 high ($46) it cost ₩14–17T. The right posture toward this stock is therefore cyclical, not directional: the utility is the asset; the pyramid is a free option only at prices near the December low. At $30 you are paying retail price for the option.

Why not a short: the floor is real. A ~4.1% gross dividend yield (restored ₩3,320/yr run-rate) on a state-protected oligopoly with ~1.6× net leverage, A3/A- credit, a 12.7% normalized FCF yield, breach recovery confirmed in Q1-2026 (+210K net adds, dividend resumed), Value-up program flows, a possible KKR ~₩1T funding validation due by ~2026-08-10 — and an AI narrative that already proved it can take this stock up +193% in five months. Short interest is a negligible 1.35% of float, but shorting a utility-with-a-story in a theme market is how you get squeezed by a press release. The factor model files SKM as a low-beta EM yield asset (USDollar −0.54 dominant loading, Korea +0.32, beta 0.42; nearest neighbors are EM bond and dividend ETFs) — the June tape priced it as an AI stock the factor model never saw, and the −33% six-week crack is the market conceding the point. The trade framing is event-driven / special-situation on the AI-premium cycle, layered on a low-beta EM yield asset — contrarian-on-pullback, never chase.

Conviction: medium. Flips bullish: AIDC margin disclosure plus contracted-revenue conversion (KKR close, Ulsan 41MW on time Nov-2027, OpenAI MoU → committed capacity) that makes the premium earned rather than hoped. Flips bearish: the post-breach run-rate proving below ₩1.7T at the Q2-2026 print, or SKT taking the 15GW vision onto its own balance sheet — the path that puts the dividend at risk again.

“Pay for the wires; get the pyramid free — not the other way around.”

📈 Stock Price Action — Five-Year Event Map

Data: AZI split/dividend-adjusted daily OHLCV for SKM (ADR), pulled 2026-07-19; FactorsToday factor model, 2026-07-17. Prices are ADR US$ closes. Price moves are FACT; attributed causes are INTERPRETATION (labeled). Factual price history only — no recommendation, no price target.

The arc (FACT): over five years SKM fell to a $15.69 low (2022-10-12, rate shock + KRW collapse), ground sideways through 2023–24, absorbed the April-2025 USIM cyber-breach, then ran +193% off the low to a $46.00 close / $47.18 intraday peak (2026-06-02) on Korea’s AI melt-up, before cracking −33.4% in six weeks to $30.62 (2026-07-17). 52-week range $19.68–$46.00; currently −33.4% off the high, +49.1% YTD, ~1% above the 200-day EMA but below the 21/50-day EMAs, on ~5× year-ago average volume. Short interest ~1.35% of float.

Event notes (move = FACT; attribution = INTERPRETATION):

1. SK Square spin-off (Nov 2021): SKT carved out its investment arm — which held the 20.1% SK Hynix stake — as separately-listed SK Square on 2021-11-29; the ADR dropped ~11% over two sessions as the distribution went ex, into a global omicron selloff.
2. 2022 drawdown: a plain-vanilla global de-rating — Fed hikes, a collapsing won (USDKRW to ~1,440), and dead-money telecom sentiment took the ADR to its five-year low of $15.69; no company-specific break.
3. USIM breach (Apr 2025): SKT detected a hack Apr 18, publicly acknowledged it ~Apr 19–22, and on Apr 28 announced free USIM replacements for all ~23M subscribers; the stock fell −9.3% on the week, its worst 2025 leg, as replacement costs, churn, and liability were priced.
4. Tariff shock (Jul 7, 2025): the US announced a 25% tariff on Korean imports; the ADR’s −7.8% was a Korea/FX-wide move with the breach overhang still on, not new company news.
5. Fine & grind (Aug–Dec 2025): the privacy regulator fined SKT ₩134.8B (Aug 28) and a government-backed consumer agency ruled for ₩100,000 per-user compensation (Dec 21); the stock ground to its 52-week low ($19.68, Dec 18) as FY25 results later confirmed op profit −41% / net −73% on hack costs — capping a year in which KRX-listed SKT fell −3.1% while the KOSPI rose +75.6% (Asiae, 2026-01-22).
6. January 2026 turnaround: the market re-rated the post-hack recovery — Anthropic-stake value in the SK orbit, 5G SA premium-pricing hopes, and one-time-cost rollover drove +34% in two weeks (KRX 52-wk high ₩63,000 on Jan 20) as sell-side flagged 2026 earnings normalization and ~6% dividend yields (third-party color).
7. AI melt-up (Feb–Jun 2, 2026): SKT’s “AI pyramid” became a price story — MWC “AI Native” strategy (Mar 2), Q1 AIDC revenue +89.3% YoY (May 7), and a NVIDIA gigawatt-scale AI-cloud partnership (Jun 7, Jensen Huang at SK HQ Jun 8) inside a KOSPI AI melt-up; the ADR spiked +19% on Jun 1 on ~16× normal volume and peaked at $46.00/$47.18 on Jun 2 (KRX ₩115,700 intraday Jun 8).
8. The crack (Jun 3–Jul 17, 2026): a US AI-chip bellwether’s soft guidance broke the Korea AI trade; the KOSPI fell −9.99% on Jun 23 (circuit breaker) and SKM — now trading as Korea/AI beta, not as a defensive telecom — gave back a third of its value in six weeks despite SKT’s announcement of a 15GW AIDC buildout (fair disclosure 6-K 2026-06-30; press early July).

Positioning note (INTERPRETATION): even after the AI rally, the factor model files SKM as a low-beta KRW/Korea yield proxy (USDollar −0.54, Korea +0.32, DividendYield +0.06; nearest neighbors are EM bond and dividend ETFs, not AI stocks), with ~81% idiosyncratic variance and a 5-year Sharpe of 0.11. The tape says “narrative round-trip, momentum broken, long trend intact at the 200-EMA” — a timing/framing input, not a verdict on the business.

1. Executive Summary

SK Telecom is Korea’s dominant telecommunications carrier — #1 in wireless for roughly forty years, with ~47% share of a closed, three-license mobile oligopoly (SKT ~47% / KT ~32% / LG U+ ~21%) that the state designed and polices. Around the wireless core sit SK Broadband (fixed-line and B tv IPTV, now wholly owned), a T-commerce stub being wound down, and the “AI pyramid” pivot: AI data centers (AIDC), GPU-as-a-service, and enterprise AI. Consolidated FY2025 revenue was ₩17,099.2B (~$11.7B), operating profit ₩1,048.5B (IFRS basis), net income attributable to owners ₩408.4B — the worst earnings year in the company’s history, for one reason: the April-2025 USIM cyber-breach.

The breach is the defining event of this report, and its resolution is the first pillar of the analysis. The direct costs are now quantified and largely behind the company: a record ₩134.8B privacy-regulator fine (paid, appeal filed January 2026), a ~₩500B customer appreciation package, free USIM replacements for ~23M subscribers, ~0.73–0.8M subscribers lost, and a ~₩700B five-year security investment commitment. FY2025 operating profit fell 38%, net income 67%, and two dividend payments were suspended. But the competitive damage was small and partially reversed — when KT suffered its own breach in late 2025, two-thirds of the subscribers fleeing KT ported to SKT. Q1-2026 confirmed recovery: operating profit ₩537.6B (+351% sequentially off the Q4 trough), +210K handset net adds, and the quarterly dividend resumed at ₩830. The breach stress-tested the moat, and the moat — positional and regulation-conferred, not brand-based — held.

The second pillar is what that moat is actually worth. SKT passes Greenwald’s share-stability test emphatically (share movement <2pp over any 5–8-year window) and fails the ROIC test just as clearly (5.4–6.1% in 2022–24; 2.2% in the breach year). The same state that blocks entry caps the rent: MSIT tariff approval, documented price interventions, subsidy caps, license obligations. This is a regulated-oligopoly moat — durable for decades, capped at utility-level returns. The no-growth earnings power of that regulated pool is worth ₩9.0–11.3T of equity (ADR $16–20).

The third pillar is the AI pivot and what the market is paying for it. AIDC revenue is real and growing fast — ₩519.9B in 2025 (+34.9%), ₩131.4B in Q1-2026 (+89.3%) — but it is ~4% of consolidated revenue, margins are undisclosed, and the market at $30.62 is embedding ₩5.8–8.2T (~$4.0–5.6B, 34–48% of equity value) of AI option value on top of the telecom EPV. Justifying that premium requires full delivery of management’s ₩5T-by-2030 AI revenue target at 20–25% EBITDA margins, into a simultaneous sector-wide Korean AI-DC build landing 2027–29. The committed slice of the program (~₩3.4T by 2028, with AWS funding the majority of Ulsan) is absorbable inside free cash flow; the 15GW national vision is categorically not SKT-fundable and is a balance-sheet tail risk, not a plan.

Governance is chaebol-standard: Chey Tae-won controls SKT through SK Inc.'s 30.6% stake (a percentage that has accreted from 26.8% in 2019 purely via SKT’s own treasury retirements, at zero cost to the holdco). The record is operationally competent — quarterly dividends since 2021, a well-timed 2023 buyback at ~₩49k, disciplined deleveraging to ~1.6× — but the hierarchy of claimants is visible: the dividend was suspended within a year of the ≥50%-payout Value-up pledge the moment breach costs and AI capex arrived, SK Broadband paid the holdco ₩506.8B cash for the Pangyo data center, and ~₩1.14T is committed into a SK Hynix affiliate. The 2021 SK Square spin-off took the 20.1% SK Hynix stake — now worth ~₩273T — out of the entity minorities were left holding.

Scenario zones (2–3-year view, ranges not targets): Bear ADR $14–20 (telecom-only EPV; the December 2025 low is the zero-premium anchor), Base ADR $27–31 (EPS restored to ₩5,500–5,800 at 11–13× plus ~₩2T of AIDC credit — the current price sits at the top of this zone), Bull ADR $40–52 (AIDC executes to plan, 14–16×). For ADR holders the won is a separate ±~9% swing factor (52-week USDKRW band 1,322–1,588) on top of anything the business does.

2. Business Overview

What the company is (FACT). SK Telecom Co., Ltd. is Korea’s largest telecommunications carrier, founded in 1984 (the successor to Korea Mobile Telecom) and listed on the NYSE as an ADR since June 1996. It is a Korean foreign private issuer reporting in KRW under K-IFRS (with the 20-F reconciled to IFRS-IASB), fiscal year ending December 31. The company operates three reportable segments — Cellular (wireless), Fixed-line (SK Broadband), and Others (T-commerce) — plus an unsegmented but strategically dominant “AI pyramid” initiative spanning AI data centers (AIDC), GPU-as-a-service (GPUaaS), enterprise AI/cloud (AIX), and consumer AI agents.

Segment economics (FACT; K-IFRS segment basis, external revenue / operating profit, KRW B; FY2025 20-F Item 5.A):

Cellular (~73% of external revenue) is the Korean mobile business: ~23.5M subscribers at YE2025 including MVNO lines (41.6% market share, still #1), postpaid-weighted, with monthly ARPU of ₩27,845 in FY2025 (down 5.1% on the breach-year appreciation package). Wireless service revenue was ₩9,715.6B in FY2025 (−6.6% YoY), the direct P&L signature of the breach response. The business model is classic facilities-based telecom: enormous sunk fixed cost (spectrum licenses — ₩664.5B of frequency rights sit in intangibles — radio access network, fiber), near-zero marginal cost per gigabyte, revenue collected as monthly subscriptions. Revenue is >95% recurring in character; churn, ARPU mix, and device-sale economics drive the variance.

Fixed-line (~25%) is SK Broadband — broadband internet, B tv IPTV, and B2B/infrastructure services. Broadband + media revenue has been flat for three years (₩2,494.0B → ₩2,510.3B → ₩2,508.1B); the growth pocket is fixed-line miscellaneous (B2B/infra, where data-center revenue largely lives): ₩1,079.6B → ₩1,180.9B → ₩1,327.2B, +12.4% in FY2025. SK Broadband became wholly owned in two steps: the ₩1,145.9B buyout of the 24.76% PE minority (agreed November 2024, paid May 2025) and a small-scale share exchange for the remaining ~0.9% completed 2026-05-29. It is also the group’s AIDC vehicle — it bought the Pangyo data-center business from SK Inc. for ₩506.8B in July 2025 (see Section 7).

Others is the T-commerce remnant, shrinking and roughly break-even after the 2025 divestitures of the SK stoa / Media S commerce subsidiaries.

Media and streaming. SKT’s real media asset is B tv IPTV — a mature, bundling-driven pay-TV product whose strategic role is churn reduction in the mobile bundle, not profit. The Wavve streaming JV (~11% Korean SVOD share, behind Netflix 31%, Tving 16%, Coupang Play 13%; Omdia via Media Play News, 2025-06-10) sits under SK Square plus the terrestrial broadcasters, not under SKM, and is being merged into a CJ ENM-controlled Tving entity (MLex, 2024-11-28) — SK would be a minority partner in a CJ-controlled company.

The AI pyramid (FACT, milestones; economics undetermined — the subject of Section 5, Section 7, Section 10):

• AIDC revenue ₩519.9B in FY2025 (+34.9% YoY); ₩131.4B in Q1-2026 (+89.3% YoY), on Gasan/Yangju utilization, the Pangyo acquisition, and the GPUaaS ramp (TelecomTV, 2026-02-06; SKT Newsroom, 2026-05-07). AIX (enterprise AI/cloud/SI) added ₩198.6B (+6.4%). Combined AI revenue ≈ ₩0.7T ≈ ~4% of consolidated revenue.
• Gasan DC GPUaaS launched January 2025 (Lambda partnership, NVIDIA H200-class hardware); the Haein cluster (>1,000 NVIDIA B200 GPUs, August 2025) was selected for the MSIT Sovereign AI project.
• Ulsan AIDC with AWS — the flagship: ~₩7T total project (~$5.1B) with AWS expected to fund ~$4B; 103MW initial phase, ~60,000 GPUs; construction started September 2025; 41MW by November 2027; full initial capacity February 2029; announced path to 1GW / ~500K GPUs. Sited near SK E&S LNG power generation — power is the binding input in Korea and the group owns it.
• OpenAI “Stargate Korea” LoI/MoU (October 1, 2025): SKT to develop/build/operate AI data centers for OpenAI in Korea, two facilities starting at 20MW; SK Hynix supplies HBM. MoU-stage — no committed capital disclosed (SK Group / KED Global, 2025-10-01).
• Equity stakes: Penguin Solutions $200M convertible preferred (December 2024, 10.3% as-converted, one board seat — filed by SKT on Schedule 13D), Lambda $20M, Perplexity $10M; Rebellions (AI chip, ex-SAPEON) consolidated.
• 15GW national AIDC vision (fair disclosure 6-K 2026-06-30): Phase 1 = 5GW opening from 2029, +10GW from 2035, “funding via strategic partners including global big-tech and overseas investors,” size/timing TBD. A vision statement, not a capital program.
• Samsung 6G AI-RAN MoU (November 2025); NVIDIA “AI factory” planned for 2027 (clarified 2026-06-08 as not gigawatt-scale at first); a KKR ~₩1T equity-investment rumor is pending a follow-up disclosure due by ~2026-08-10 (6-K 2026-07-10).

Business-model verdict. SKT is a subscription utility with a call option attached. The core earns regulated-oligopoly rents — stable, non-discretionary demand, politically capped pricing — and converts them to cash with high fidelity (operating cash flow has held ₩3.9–5.8T every year for a decade). The AI layer is a genuine second business in formation but is today 4% of revenue with undisclosed margins; it does not yet change what the company is. Any analysis that starts from the pyramid rather than the wires is starting from 4% of the business.

3. Industry Dynamics

3.1 Structure: a closed, state-designed oligopoly

Korean mobile is a three-carrier facilities oligopoly — SKT ~47%, KT ~32%, LG U+ ~21% (2025 estimates; matrixbcg, 2026-03-31) — plus a government-supported MVNO (“altno”) tail that is far weaker than the US cable-MVNO leak. There is no cable competitor and no fixed-wireless bypass story of consequence. Entry is effectively impossible: spectrum is assigned by the Ministry of Science and ICT (MSIT) to the three incumbents, and repeated attempts to license a fourth MNO have been rejected. The oligopoly is state-designed, not market-evolved — the crucial fact for locating the source of any competitive advantage. The three players coordinate visibly when it suits the state, most strikingly the simultaneous April-2019 5G launch following the June-2018 3.5GHz/28GHz auction (TeleGeography, 2018-06-19).

3.2 Regulation: the state is the fourth player

The regulatory apparatus is not background noise; it is the industry’s defining economic feature (FACT, documented interventions):

• Tariff approval. SKT, as the designated dominant carrier, must obtain MSIT approval for its rate plans (KT and LG U+ merely file) — a formal price-ceiling mechanism with no US analogue.
• Documented price interventions. 2017: all three carriers raised the optional tariff discount from 20% to 25% “in compliance with policy initiatives announced by the MSIT” (KT 20-F, SEC). 2019: “at the urging of the MSIT, all three companies offered a very basic 5G plan for KRW55,000/month” (Display Daily, 2019-09-18).
• Subsidy caps. The 2014 Handset Distribution Reform Act capped device subsidies and ended the subsidy wars — the state deliberately suppressed the one competitive lever carriers had, freezing the share structure in place.
• License obligations. Coverage and build commitments attach to spectrum grants; the state excludes the MNOs from the private-5G operator market on anti-monopoly grounds (NetManias, 2025-09-18) — a market fenced off from the oligopoly.
• Security and consumer punishment. The 2025 breach response showed the state’s enforcement toolkit: mandatory ETF waivers for departing customers (deliberately lowering switching costs as punishment), a record ₩134.8B PIPC fine, and a seven-week suspension of new subscriptions. New Korean legislation (March 2026) raises CISO accountability and surcharges for repeat incidents.
• 6G. MSIT wants commercialization by 2029; the operators are publicly urging a slower, hotspot-focused approach, citing poor 5G returns (DIGITIMES Asia, 2024-07-22). The carriers now push back — a shift from the compliant 5G era — but the state retains the spectrum lever.

3.3 Demand: saturated, 5G-mature, shrinking

5G reached 33.85M subscribers — 65.4% of the population — by May 2024, with nationwide coverage achieved April 2024 and monthly net-add growth down to ~0.7% (CMS, 2025-03-04): the migration is complete. Total Korean telecom + pay-TV service revenue is forecast at a ~1.9% CAGR for 2024–2029, with mobile data at ~5.6% (GlobalData via GlobeNewswire, 2025-01-20). Korea’s population peaked in 2020 (~51.8M) and is declining with the world’s lowest fertility rate (~0.75 in 2024, Statistics Korea). The subscriber base is more than fully penetrated and the addressable population shrinks every year: volume growth is over; the game is ARPU and mix under a political price ceiling.

3.4 Korea vs. US telecom economics

Prior analysis of the US carriers (AT&T, Verizon, T-Mobile US, from their public filings) concluded the US wireless industry is “structurally good but mature, historically value-destructive, bifurcated capital cycle.” Korea is the same architecture with lower returns:

The US verdict applies to Korea with lower returns: the same fortress entry barriers, but the state captures a larger share of the oligopoly rent on behalf of consumers.

3.5 Capital-cycle read (Marathon)

The core telecom cycle is turning favorable: the 5G build is complete (nationwide coverage April 2024), SKT capex fell from ₩3.69T (2020) to ₩2.32T (2025) with wireless-network capex nearly halved to ₩733.9B in FY2025, no fourth entrant exists, shares are stable, and carriers are resisting the 6G capex recall. Falling investment against stable, non-discretionary demand is the textbook setup for improving forward returns on the core. But Marathon’s own caveat governs: regulation flatlines the cycle — returns never boom (the state caps price) and never bust (the state blocks entry) — so the cycle mean-reverts to a regulated level, not a competitive one. Meanwhile AIDC is a new capital-inflow wave at the boom stage: all three telcos, Naver, Samsung SDS, LG CNS, the government, and OpenAI/SoftBank capital are simultaneously entering Korean AI infrastructure, with capacity landing 2027–29 — the same supply-side flag raised by the memory sector’s simultaneous capex wave (public company disclosures).

Verdict (Section 7.2): a structurally stable, low-growth, returns-capped industry — “good” on durability, “mediocre” on economics. Entry barriers are near-absolute and demand is non-discretionary, so cash flows are more predictable than in most industries. But pricing is politically capped, the population is shrinking, and every technology cycle is a state-sponsored capital call. The profit pool sits with the three carriers in stable 47/32/21 proportions, with the government as a perpetual marginal claimant (tariff approval, subsidy caps, fines, license obligations). The industry favors scale incumbents by design; it does not let them get rich.

4. Competitive Position

4.1 The Greenwald tests, run honestly

Share-stability test: PASSES — emphatically. SKT has been Korea’s #1 carrier since the Korea Mobile Telecom era — roughly forty years — holding ~47–48% share for decades, with only low-single-digit drift to LG U+ over 2014–2019 (GlobeNewswire, 2020-09-01). Share movement of <2pp over any 5–8-year window clears the framework’s “formidable barriers” threshold by a wide margin.

ROIC test: FAILS — just as clearly. ROIC ran 5.37% / 5.87% / 6.14% in 2022/2023/2024 and 2.21% in the 2025 breach year (ROIC.ai aggregated data, 2026-07-19). Greenwald’s taxonomy reads sustained 15–25% ROIC as evidence of genuine franchise advantages and 6–8% as their absence. SKM sits at utility-level returns — below any reasonable estimate of its cost of capital’s equity component, around a 7–8% WACC.

Reconciliation (INTERPRETATION — the honest answer, and the spine of this report): the moat is real as a stability mechanism and absent as a super-normal-returns mechanism, because regulation confers both the protection and the cap. The state grants a three-license franchise and then prices the consumer benefit out of the carriers’ P&L through tariff approval, subsidy caps, and license obligations. In Greenwald’s terms this is a regulated-oligopoly moat: economies of scale plus light customer captivity, with the returns ceiling set politically. The barrier is entry-blocking — three licenses, assigned spectrum, multi-trillion-won sunk network costs — so the position is durable for decades; the returns are capped because the same state that blocks entry caps price.

4.2 Switching costs: real but modest, and state-eroded

Number portability has existed since 2004; the pre-2014 subsidy wars made switching lucrative until the Handset Reform Act capped subsidies and chilled it. What remains is habit and bundle friction: family/combined discounts across mobile + B tv IPTV + broadband, T-membership loyalty, and inertia. The breach proved customers can leave quickly when the state lowers the drawbridge — ~166K subscribers in the 10-day ETF-waiver window, ~800K total — but also that they rotate among the same three carriers, and that the flow reverses when a rival stumbles (Section 4.4). Scale advantages in procurement (handsets, network gear), network opex per subscriber, and brand/media assets are real but shared in kind with KT and LG U+ at smaller scale — a cost advantage of degree, not kind.

4.3 What would deteriorate without the position

The ~₩1.7–1.8T/yr operating-profit pool is the regulated oligopoly rent. Without the license-protected #1 scale position, the financial signature to expect: ARPU compression toward MVNO levels, churn-driven marketing costs up, operating margin from ~10% toward low single digits, ROIC structurally below WACC. Conversely, the financial signature of the moat as it exists is the stability of the 47% share and the ~10% operating margin through a full technology cycle — not the level of either. The 2025 breach supplied exactly the stress test the framework asks for, and the stability signature held.

4.4 The breach as a moat test (FACT, cited; the decisive competitive datapoint)

The full timeline is in Section 8; the competitive reading is here. When the MSIT probe found SKT at fault (28 of 42,605 servers infected, 33 malware types, plaintext credentials, unpatched systems, ignored logs — Light Reading, 2025-07-07), the state ordered ETF waivers for departing customers — deliberately lowering switching costs as punishment. SKT lost ~0.73M subscribers in the April 19–July 14 window and ~800K in total (TelecomTV, 2026-02-06); the feared worst case of 2.5M (Telecom Magazine, 2025-07-04) did not materialize. Then the decisive event: KT suffered its own breach (September 2025, unauthorized micropayments), and when KT waived ETFs on December 31, 2025, it lost 216K subscribers in under two weeks — faster than SKT had — and two-thirds of the January 10 outflow (22,193 of 33,305) ported TO SKT (Light Reading, 2026-01-12; KoreaTechToday, 2026-02-20). Q1-2026: SKT posted +210K handset net adds.

Assessment (INTERPRETATION): the breach was a one-time financial impairment of ~₩1T-class all-in with a small, largely reversed competitive cost — not a structural impairment of the franchise. Three reasons: (1) the oligopoly is a closed loop — churned customers rotate among the same three carriers, and SKT recaptured share from KT’s own incident; (2) the state’s punishment mechanism (the ETF waiver) expired; (3) brand/trust damage proved temporary relative to rivals carrying the same risk. Residual structural costs are real — permanently higher security opex (the ~₩700B five-year plan), industry-wide regulatory tightening, pending civil suits — but they tax the whole oligopoly, not SKT’s position within it. The moat survived its stress test, which is itself evidence the moat is positional/regulatory rather than brand-based. If the franchise ran on brand trust, April 2025 would have broken it.

Verdict (Section 7.3): durable but capped competitive advantage — a regulation-conferred oligopoly franchise plus genuine scale, NOT a Greenwald franchise business. It passes the share-stability test and fails the ROIC test; it is a franchise for stability, not for returns. The disconfirming evidence has been weighed: a brand-based moat would have shown larger, persistent share loss in 2025 (it didn’t); a scale-economics moat in the Greenwald sense would show 15%+ ROIC (it shows 5–6%). What remains is exactly what the industry structure predicts: unassailable position, regulated returns.

5. Growth History and Forward Opportunities

5.1 Historical growth: there isn’t any, and that is the point

Consolidated revenue (KRW B, IFRS): 16,748.6 (FY2021) → 17,305.0 (FY2022) → 17,608.5 (FY2023) → 17,940.6 (FY2024) → 17,099.2 (FY2025, breach year). That is a ~1.8% CAGR through FY2024 against Korean CPI, i.e., approximately zero real growth — and FY2025 revenue is barely above the restated FY2019 continuing-operations level. Segment detail confirms the stagnation is structural, not cyclical: wireless service revenue peaked at ₩10,401.6B (FY2024) before the breach took it to ₩9,715.6B; broadband + media has been flat at ~₩2.5T for three years; the commerce stub shrank by a third and is being exited. The only line growing double digits is fixed-line miscellaneous (B2B/infra — where data-center revenue lives), +12.4% to ₩1,327.2B in FY2025.

The growth record decomposes into: (a) subscriber volume — finished (>100% penetration, shrinking population); (b) ARPU — capped (MSIT tariff approval; the FY2025 decline to ₩27,845 was breach-driven, but the pre-breach trend was low-single-digit at best); © mix — mildly positive (5G at 65.4% penetration, premium plans, 5G SA optionality); (d) acquisitions — episodic (SK Broadband consolidation, Pangyo DC). Organic growth in the core is a 1–2% affair, and the state’s price ceiling means even that must be defended politically.

5.2 Forward opportunities, ranked by evidence

1. AIDC / AI infrastructure — the only real growth engine (FACT on revenue, OPEN QUESTION on returns). AIDC revenue ₩519.9B (FY2025, +34.9%) → ₩131.4B (Q1-2026, +89.3%); AIX ₩198.6B (+6.4%). Management targets AI revenue of at least ₩5T/yr by 2030 (September 2025 fair disclosure; the October 2024 Value-up plan had framed AI as 35% of a ₩30T 2030 revenue ambition), with AIDC revenue alone targeted above ₩1T by 2030 (company commentary). Against ₩519.9B of FY2025 AIDC revenue, the broad ₩5T target implies roughly 7–10× growth in five years. The demand side has genuine substance: state-backed Korean AI-computing sovereignty programs, corporate demand (Samsung, Naver, startups), power-constrained Seoul-metro capacity scarcity, and SK Group input advantages (SK E&S LNG power at Ulsan, SK Hynix HBM adjacency, SK ecoplant construction). AWS anchoring Ulsan is demand validation, not speculation. The caution side is equally concrete: every Korean competitor is building simultaneously (Naver GAK Sejong, KT Cloud +27.4% in 2025 with a Microsoft partnership, LG CNS, Samsung SDS, the state’s own national AI computing project, plus hyperscalers); GPUaaS is commodity compute rental with no captivity and GPU-price-dependent margins; the OpenAI “Stargate Korea” agreement is MoU-stage with no committed capital; and management claims AIDC profitability is “already comparable to telecom” while refusing to disclose the number (Q1-2026 call, CFO Park Jong-seok). Industry synthesis: ~60% real economics, ~40% chaebol re-rating narrative — real demand and real group synergies, wrapped in a strategic motive that plainly includes re-rating a 5%-ROIC telecom as an “AI company.”

2. 6G — an obligation more than an opportunity. MSIT targets 2029 commercialization; the carriers are publicly resisting a full build, citing poor 5G returns (DIGITIMES Asia, 2024-07-22). The Samsung AI-RAN MoU (November 2025) is positioning. Treat 6G as a future capital call with regulated recovery, not a growth vector.

3. 5G SA / ARPU mix and post-breach recapture. Q1-2026’s +210K net adds and the KT-breach inflow show near-term ARPU and share recovery is achievable; sell-side January-2026 enthusiasm for 5G SA premium pricing is plausible but unproven (third-party color).

4. Media. Wavve (11% SVOD share) sits under SK Square and is merging into CJ-controlled Tving; B tv is a bundling tool. Media is a churn-reduction cost center, not a growth asset.

Verdict (Section 7.4): low-quality growth at the core, one genuinely uncertain growth option at the edge. The core grows 1–2% nominal with a political ceiling on the only lever (price) that could change that — this is low-quality growth by definition, stable rather than expanding. AIDC is high-velocity growth (35–89%) off a 4%-of-revenue base into a crowded simultaneous build, with the single most important number (margins) undisclosed. Growth without demonstrated economics is not investable growth; the economics question is deferred to Section 10, where the market’s price for it is reverse-engineered.

6. Financial Quality

6.1 Multi-year summary (FACT; KRW B; op profit IFRS per 20-F; EPS basic per ORDINARY share, ₩)

* FY2021 NI includes ₩1,147.6B of discontinued-operations gains (SK Square spin-off, effective 2021-11-01). Capex = cash purchases of PP&E + intangibles; the 20-F’s PP&E-basis capex table reads 2,973.9 (FY23) → 2,487.4 (FY24) → 2,206.6 (FY25). Revenue, NI, OCF, capex, and dividends tie exactly to EDGAR XBRL for all years.

Quarterly anatomy of the breach (ROIC, KRW B): Q1-25 rev 4,453.7 / op 574.9 (pre-breach); Q2-25 4,338.8 / 338.3; Q3-25 3,978.1 / 61.2 — the trough, net loss ₩166.7B (controlling −158.2B) as the fine and appreciation package concentrated; Q4-25 4,328.7 / 28.2 (K-IFRS ₩119.1B); Q1-26 4,392.3 / 547.8 (K-IFRS ₩537.6B, +351% QoQ, −5.3% YoY), NI controlling ₩322.4B — management declared “return to pre-incident performance” (Q1-26 earnings 6-K, 2026-05-07; call summary). Q1-26 OCF was 1,064.2 (vs 1,252.0 in Q1-25) with capex of only 382.7.

6.2 Breach cost accounting (FACT; where each won landed)

The K-IFRS/IFRS presentation nuance matters for anyone cross-checking Korean press: SKT’s headline domestic operating profit (₩1,073.2B FY2025) excludes the ₩134.8B fine, which sits below the K-IFRS operating line in other expenses; the IFRS figure (₩1,048.5B) includes it (20-F p.55 reconciliation). The fine’s non-deductibility also drove the FY2025 effective tax rate to 48.1% versus a normal 21–23%.

6.3 Normalized vs. reported (INTERPRETATION, built on FACT strips)

Stripping only the fine (~₩135B) and the appreciation package (~₩500B) from FY2025 yields an adjusted net income of roughly ₩950B–1,000B; a fuller normalization (churn revenue recovery, ETR back to ~23%, no repeat of the Q3 trough) supports the working normalized base used throughout this report: operating profit ~₩1,700B, net income ~₩1,200B, EPS ~₩5,500–5,800 per ordinary share, EBITDA ~₩5,300B (op ~1,700 + D&A ~3,600; FY2025 D&A was 3,555.6). The open question — flagged, not resolved — is that Q1-2026’s ₩537.6B annualizes to ~₩2.15T, above the ₩1.7T normalized base; some of that is KT-breach share recapture that may fade. If the true post-breach pool is ₩1.9–2.0T, every valuation zone in Section 10 shifts up ~₩6,000–8,000 per ordinary share. Q2-2026 (early August) is the test.

6.4 Balance sheet and returns (FACT, 2025-12-31 unless noted; KRW B)

Assets ₩30,107.8; cash ₩1,490.0 plus short-term investments ₩186.6; LT investments ₩5,427.4 (associates/JVs + AI stakes); net PP&E ₩11,902.2; goodwill ₩2,072.5; intangibles ₩1,710.6 (incl. ₩664.5 frequency rights). Gross debt ₩10,372.8 (debentures 8,213.9 + borrowings 633.1 + lease liabilities 1,525.8; 16.2% FX-denominated, mostly USD) against cash+STI of 1,676.6 → net debt ₩7,357.0 (₩7,308.8 at Q1-26). Equity attributable ₩12,863.1 (including ₩398.5B of 2023 hybrid bonds, equity-classified, ~₩19.8B/yr interest); NCI ₩92.2. Net debt/equity 56.8%.

Leverage on the corrected basis: net debt / true EBITDA ≈ 1.60× (FY2025 breach-depressed EBITDA ≈ ₩4,604B = IFRS op 1,048.5 + D&A 3,555.6; FY2024 ≈ 1.28×). Interest expense ₩383B covers ~12× on EBITDA. Ratings: Moody’s A3 stable (affirmed April 2025, mid-breach), S&P A−, Fitch A−, domestic AAA. Returns: ROE 1.78% (FY25) / 5.45% / 4.83% / 4.06% (FY24/23/22); ROIC 2.21% / 6.14% / 5.87% / 5.37%. Even normalized, ROE ~8–9% and ROIC ~5–6% — the financial fingerprint of the returns ceiling discussed in Section 4.

6.5 Data-quality traps (FACT — documented so no downstream reader inherits them)

The ROIC.ai aggregation layer is materially wrong for SKM in four specific ways, all verified against the 20-F/EDGAR: (1) its FY2025 “EBITDA” line equals operating income (D&A not added back) — producing a nonsensical 7.25× net-debt/EBITDA and 19.9× EV/EBITDA versus true ~1.6× / ~4.4–5.3×; (2) its “FCF yield 54.7%” is a sign error (its TTM FCF field adds capex to OCF: 3,923.8 + 2,323.3) — true FY2025 FCF is ₩1,600.6B, a ~12.7% yield on normalized basis at the current cap; (3) its per-share fields (EPS/DPS/BVPS/FCF-per-share) are per ADS in KRW on a ~383.7M ADS-equivalent count — multiply by 1.8 for per-ordinary-share KRW (true FY2025 basic EPS is ₩1,825 per ordinary share; ROIC shows ₩1,064 per ADS, also skipping the ₩19.8B hybrid-bond interest deduction); (4) its EV is keyed to a stale YE2025 price (~₩11.4T market cap) — the live market cap is ~₩17.1T. Every multiple in this report is rebuilt by hand from filing-tied figures and the live price.

6.6 Verdict (Section 7.5): the economics do not improve with scale — they merely persist

SKT is a genuine cash-generative incumbent: OCF has held ₩3.9–5.8T every year 2016–2025, and FCF stayed positive (₩1.6T) even in the worst earnings year in company history. But it is a zero-operating-leverage business — FY2025 revenue is barely above restated FY2019, operating margin is capped at ~9–10% by MSIT rate intervention, and ROIC of 2–6% sits at or below any reasonable WACC. The breach demonstrated how thin the earnings cushion is: one incident consumed 63% of net income and two dividend payments. The AIDC pivot is real on milestones but invisible in the financials so far — consolidated capex is falling (wireless capex nearly halved to ₩733.9B in FY2025) and no AIDC segment exists; the gigawatt narrative is funded capacity-aspiration, not current cash flow. The balance sheet (~1.6× true leverage, AAA domestic) can fund the committed build, but FY2026+ guidance explicitly contemplates “additional AI-related investments” — watch for FCF compressing as Ulsan construction peaks toward 2027 operations. Disconfirming evidence weighed: capex discipline in the core is genuinely shareholder-favorable (the Marathon setup of Section 3.5), and Q1-2026’s +351% sequential recovery says the earnings base is intact. Neither changes the verdict: this is a utility-grade P&L, and it should be owned — if owned — as one.

7. Capital Allocation

7.1 The control chain (FACT; FY2025 20-F Items 6/7)

Chey Tae-won (chairman) → SK Inc. (group holdco) → 30.6% of SKT (65,668,397 shares; 30.8% of outstanding). Two facts define the governance economics. First, the percentage accreted for free: SK Inc.'s share count has been unchanged since the 2021 spin-off, but its percentage rose from 26.8% (2019–20) → 30.0% (2021) → 30.6% (2024–25) purely through SKT’s own treasury-share retirements — mechanical accretion of control with zero cash spent by the holdco. Second, minority-relevant owners are exiting or absent: the National Pension Service has been a persistent multi-year seller (11.0% in 2019 → 6.7% now); directors and executive officers together hold <0.1% (172,012 shares); the new CEO holds 1,518 shares. There is no founder-family stake at the SKT level and no dual class — one-share-one-vote inside a chaebol chain. Foreign ownership is ~36% versus the 49% statutory cap (Telecommunications Business Act; SK Inc. deemed domestic — a change-of-control tripwire exists if SK Inc.'s own register turns foreign, currently not the case). Wellington crossed 5% (13G, 2026-05-15, passive). As an FPI there is no Section 16 insider-transaction stream to read; ownership structure and the 6-K corporate-action record are the substitutes.

7.2 The SK Square spin-off (2021): mechanically fair, economically a value migration

On November 1, 2021, SKT executed a horizontal spin-off: the 20.1% SK Hynix stake plus security (ADT Caps, SK Infosec), commerce (11st), T Map Mobility, and SK planet went to SK Square; SKT shareholders received SK Square shares pro rata; SKT’s share count was cut 39.26% and ₩14.82T of book equity left the balance sheet (LT investments ₩16.0T → ₩3.9T). Mechanically pro-rata-fair — a 2021 SKT holder who kept both legs owns the Hynix windfall. Economically, weigh what the value became: SK Square’s stated NAV at spin was ~₩26T; today SK Hynix trades at ~₩1,357T market cap, making the 20.1% stake worth ~₩273T, while SK Square itself trades at ~₩160T — a ~41% holding-company discount to the Hynix stake alone. The entity minorities were left holding (SKT) lost the group’s crown jewel and its ~₩1T/yr of equity-method income (pre-spin FY2017–18 net income ran ₩2.6–3.1T on Hynix contributions), while the Chey/SK Inc. layer retained control of both companies at zero cash cost — and even saw its SKT percentage rise. Disconfirming evidence, honestly weighed: the spin was pro-rata (minorities could keep the Hynix exposure, and many did — the +193% ADR run into June 2026 partly rode the same AI complex), and pre-spin SKT’s Hynix-flattered earnings obscured the telecom’s standalone mediocrity. Neither undoes the structural read: the value migrated out of the listed operating company into a Chey-controlled vehicle that the market prices at a steep holdco discount.

7.3 Related-party flags (FACT; FY2025 20-F Item 7.B + 6-K stream)

• Pangyo data center: in July 2025 SK Broadband bought the Pangyo DC business from SK Inc. for ₩506.8B cash — a business combination under common control, priced at carrying amounts in SK Inc.‘s books. Not obviously value-destructive on price, but directionally it is the holdco monetizing an asset into SKT’s AI pivot: cash moved from SKT minorities’ consolidated entity up to the holdco in the same year the dividend was halved.
• SKHNPS / Solidigm: in June 2026 SKT committed ~₩1.14T into SK hynix NAND Product Solutions Corp. (the US-incorporated Solidigm successor, an SK Hynix affiliate) — ₩738.4B capital contribution (0.9% stake, scheduled June 2030) plus ₩397.1B share acquisition (0.62%, July 2026) — “to facilitate synergies with AI business” (6-Ks 2026-06-26/06-30). The Fair Trade Act generally restricts intra-group shareholding; the US-incorporation threads that needle. The structure — SKT committing nine figures of minority capital into a group semiconductor affiliate for tiny stakes, much of it deferred — looks more like chaebol capital recycling than arm’s-length investment (INTERPRETATION).
• Routine flows: SK Inc. brand-licensing + IT service fees ₩440.2B (2025), IT equipment purchases ₩256.3B (2025, up from ₩125.7B); SK Networks dealer commissions ₩977.9B. The 2023 buyback was executed through SK Securities, an affiliate broker.
• SK Inc. monetization overhang: the holdco has historically monetized its SKT stake via SPV exchangeable-bond structures linked to SKM ADSs (a 20-F risk factor flagging ADS short-selling pressure from such deals).

7.4 Dividends and buybacks: the empirical answer to “who gets paid first?”

Dividends (FACT, per ordinary share): quarterly system since 2021 — FY2023 ₩3,540 (830×3 + 1,050 annual); FY2024 ₩3,540; FY2025 ₩1,660 only (Q1+Q2 ₩830 each; the Q3-2025 interim AND the FY2025 year-end dividend were both suspended — the breach). Cash dividends paid: ₩773.8B / 804.3B / 628.4B (FY23/24/25) — note FY2025’s ₩628B paid was 168% of breach-year net income: the cut still out-earned the P&L. Policy: ≥50% of adjusted consolidated net income for 2024–2026 under the October 2024 Value-up plan (CFO Kim Yang-seob called the 50% “a symbolic lower bound” on the Q3-2025 call), plus ROE ≥10% by 2026 — a pledge FY2026 is tracking to miss (~8%). Q1-2026: ₩830 quarterly resumed (6-K 2026-04-27), with management “aiming to restore payout levels over time”; the March 2026 AGM approved a capital-reserve reduction, creating capacity for future returns. At the restored ₩3,320/yr run-rate, the ADR yields ~4.0–4.1% gross at $30.62 (₩1,967/ADS ≈ $1.24 at 1,460); Korean withholding (~22% statutory; 15% + local sutax under the US treaty) and depositary fees trim that, and the USD payout carries full KRW translation risk.

Buybacks (FACT): 2020 ₩426.7B; May-2021 retirement of 8.69M treasury shares (~₩2.6T announced, ₩1,966B book — the pre-spin cleanup that lifted SK Inc. to 30.0%); 2023 ₩285.5B at ~₩49.4k average (well-timed ex-post versus ~₩80k today); 2024 tail ₩15.8B plus cancellation of 4.04M treasury shares; 2025: zero; 2026 YTD: none announced. The pattern is unambiguous: buybacks are opportunistic and flow-dependent, not countercyclical — the program stopped entirely when the stock was cheapest (2025, ADR 52-week low $19.68), the exact moment a shareholder-first allocator would have been most aggressive.

7.5 The AIDC funding reality check (INTERPRETATION on FACT inputs)

Normalized OCF is ₩4.9–5.2T/yr; capex has been falling (₩3.69T 2020 → ₩2.32T 2025) as the 5G build ended; FCF runs ₩1.6–2.5T. SKT’s own committed AI-infrastructure slice — ~₩3.4T by 2028, ≈ ₩1.1T/yr — is absorbable inside FCF alongside an ~₩800B/yr dividend if AWS and partners carry ~60%+ of Ulsan (~$4B of the ~₩7T) and telecom EBITDA (₩4.5–5.5T) holds. The balance sheet (net debt ₩7.36T, ~1.6×, A3/A−/AAA) has ample capacity for the committed program. The 15GW vision is categorically not SKT-fundable (at industry costs of ~$10B+/GW it is a $50B+ class ambition against an ₩17T market cap) — it requires hyperscaler capital, project finance, and group co-investment; the risk to watch is SKT being designated the group’s funding vehicle, and the Pangyo precedent shows cash can flow either direction across the holdco boundary. The dividend interaction has already been answered empirically: in H2-2025 the dividend was the swing variable when breach costs + AI capex + the ₩1.15T SK Broadband buy-in coincided. The pyramid gets funded first; minorities are paid from what is left.

7.6 Management incentives (FACT — thin FPI disclosure, flagged)

Aggregate director comp FY2025 ₩5.3B; executive officers ₩60.4B. Named: outgoing CEO Ryu Young-sang received ₩3,407M (salary 1,540 + bonus 1,820) — a ₩1.8B bonus in the year net income fell 67% and Korea’s largest data breach occurred on his watch; he was moved up to the SK Group Supex Council in the October 30, 2025 reshuffle, not held accountable in any economic sense. New CEO Jung Jai-hun (₩2,079M from October 2025) is a governance/legal profile — a former presiding judge and SKT’s ex-Chief Governance Officer — confirmed representative director at the March 26, 2026 AGM; the CFO also changed (Kim Yang-seob → Park Jong-seok). A PSU program has replaced stock options since 2023 (3-year vesting, treasury-settled, CEO eligible up to +100% of salary) — but the performance metrics are not disclosed in the 20-F, so it cannot be verified whether PSUs pay on ROE/TSR or vanity growth. That is a disclosure gap, flagged as an Open Question in Section 13.

7.7 Verdict (Section 7.6): competent stewards of a cash machine that belongs to the chaebol first — mixed-to-negative, not minority-first

The genuinely shareholder-friendly chapters are real: quarterly dividends since 2021, a ₩2.6T treasury retirement, a well-timed ₩300B buyback at ~₩49k, the 4.04M-share cancellation, a published ≥50%-payout Value-up plan, disciplined deleveraging to ~1.3–1.6×, and falling core capex. Weigh against them: (1) the SK Square spin stripped minorities’ residual entity of the group’s best asset and ₩1T/yr of equity income while control cost SK Inc. nothing; (2) the dividend — the one binding promise to minorities — was suspended within a year of the Value-up pledge the moment breach costs and AI capex arrived, and zero buybacks were executed at the 2025 lows; (3) SKT paid the holdco ₩506.8B cash for Pangyo and ₩1,146B for SK Broadband minorities in the same year it halved the dividend; (4) the breach-year CEO collected a ₩1.8B bonus and was promoted to the group council; (5) management refuses to disclose AIDC profitability while claiming telecom-comparable margins, and the 15GW vision implies future claims on SKT’s balance sheet that the dividend would again absorb. The hierarchy of claimants, in order: national AI champion > SK Group > minorities. The 2026 dividend resumption and any AIDC returns disclosure are the proof points that could revise this verdict; neither has yet done so.

8. Changes and Headwinds — Last Two Years

The last twenty-four months contain the densest event sequence in the company’s modern history. Timeline below is FACT (6-K stream, 20-F, cited press); the synthesis is INTERPRETATION.

2024 — the setup.

• Apr 2024: Shareholder Return Policy FY2024–26 adopted — ≥50% of adjusted consolidated profit via dividends and/or buyback-cancellation. Nationwide 5G coverage achieved.
• Oct 24, 2024: Corporate Value-Up Plan published (KRX value-up initiative): ROE ≥10% by 2026; ≥50% payout; “AI Vision 2030” — ₩30T revenue with 35% from AI by 2030.
• Nov 13, 2024: Board approves acquiring the remaining 24.8% of SK Broadband from its PE minorities for ₩1,145.9B (completed May 2025) — consolidating the DC vehicle ahead of the AIDC push.
• Dec 2024: $200M Penguin Solutions convertible preferred via SPV Astra AI Infra LLC (Schedule 13D filed by SKT, 2024-12-20; 10.3% as-converted, one board seat). FY2024 closes as the best recent year: revenue ₩17,940.6B, op profit ₩1,690.9B, EPS ₩5,780 — the pre-breach baseline.

2025 — the breach year.

• Feb 12, 2025: FY2025 guidance of ₩17.8T revenue with operating-profit growth — obsolete within weeks.
• Apr 18–19, 2025: USIM cyber-intrusion detected and disclosed; malware subsequently traced to August 2021. USIM data (IMSI/IMEI/authentication keys) of ~23–27M records compromised.
• May 5 – Jun 23, 2025: new mobile subscriptions suspended (MSIT administrative guidance; USIM replacement inventory shortage) — seven weeks with the sales engine off.
• Jul 4–7, 2025: “Accountability and Commitment Program” and guidance cut (revenue ₩17.8T → ₩17.0T): free USIM replacements for ~23M subscribers, free USIM-protection service, one year of free third-party device security, August fee discounts plus free data to year-end (~₩500B Customer Appreciation Package), ETF waivers for departing customers (~0.73M subscribers lost Apr 19–Jul 14), and a ~₩700B five-year cybersecurity investment plan. The MSIT probe found SKT at fault (28 of 42,605 servers infected; plaintext credentials; ignored logs).
• Aug 27, 2025: PIPC fine ₩134.8B — the largest in the regulator’s history — plus ₩9.6M administrative penalty and a correctional order; provisionally paid; administrative lawsuit filed January 2026.
• Sep 2025: AI reorganization (AI company-in-company); fair disclosure of ₩5T cumulative AI investment over five years and AI revenue ≥₩5T by 2030. SKT rejects PIPC dispute-mediation proposal (₩300K × 3,998 claimants); KT suffers its own breach.
• Oct 30, 2025: Q3-2025 dividend skipped; Q3 preliminary results show revenue −12.2% YoY, operating profit −90.9%, net loss ₩166.7B. SK Group reshuffle replaces CEO Ryu Young-sang with Jung Jai-hun (ex-Chief Governance Officer, former presiding judge); Ryu moves to the group-level Supex Council. CFO also replaced (Park Jong-seok).
• Nov–Dec 2025: KT’s ETF waiver (Dec 31) triggers 216K KT defections in two weeks, two-thirds of the January outflow porting to SKT — closed-loop churn in action. Consumer agency proposes ₩100K × 58 claimants. The ADR grinds to its 52-week low of $19.68 (Dec 18); KRX-listed SKT finishes 2025 −3.1% against a KOSPI up +75.6%.

2026 — recovery and the AI re-rating.

• Feb 5, 2026: FY2025 preliminary results — revenue ₩17.099T (−4.7%), op profit ₩1.073T K-IFRS (−41.1%), net income attributable ₩408.4B (−67.3%) — and no year-end dividend (FY2025 total ₩1,660 vs ₩3,540).
• Mar 26, 2026: AGM confirms Jung as representative director; approves the SK Broadband share exchange (cashing out the residual ~0.9% at ₩15,032/share) and a capital-reserve reduction (return capacity); enabling tax-free dividends approved.
• Apr 27–29, 2026: quarterly dividend resumed at ₩830 for Q1-2026; FY2025 20-F filed (2026-04-29).
• May 7, 2026: Q1-2026 preliminary results — revenue ₩4.392T (−1.4%), op profit ₩537.6B (+351% QoQ, −5.3% YoY), net income ₩322.4B; AIDC revenue ₩131.4B (+89.3%); +210K handset net adds; management declares “return to pre-incident performance.” SK Broadband share exchange completes May 29 → wholly owned.
• Jun 2026: NVIDIA gigawatt-scale AI-cloud/DSX headlines (clarified Jun 8: the first 2027 “AI factory” is not gigawatt-scale; expansion in phases); ~₩1.14T SKHNPS/Solidigm related-party commitment (Jun 26/30); 15GW AIDC buildout fair disclosure (Jun 30) — 5GW from 2029, +10GW from 2035, partner-funded, size/timing TBD. KOSPI AI complex cracks (−9.99% circuit breaker Jun 23); SKM falls −33.4% in six weeks from the $46.00 Jun-2 peak.
• Jul 2026: KKR reportedly reviewing a ~₩1T equity investment in SKT’s AI-DC project — company confirms options under review, nothing decided, follow-up disclosure due by ~2026-08-10 (6-K 2026-07-10). KT’s PIPC fine decision expected July 2026 (read-through for sector-wide breach economics).

Verdict (Section 7.7): the two-year sequence strengthens the utility thesis and weakens the pyramid thesis. The breach year proved three durable things: the earnings floor is regulated and recovers fast (Q1-2026); the closed-loop oligopoly contains competitive damage (KT’s defection inflow); and the dividend is the first variable sacrificed when group priorities collide with it. The AI sequence proved two uncomfortable things: the narrative can double the stock on MoU-stage economics (Feb–Jun 2026), and every leg of the build increases SKT’s entanglement with group-adjacent funding (Pangyo, SKHNPS, the pending KKR structure). Headwinds that remain live: breach civil-litigation tail, the ~₩700B security-spend drag, MSIT’s 6G ambitions, and the 2027–29 Korean AI-DC capacity wave. Nothing in the sequence damaged the franchise; several things in it damaged the case for paying an AI premium today.

9. Risk Analysis

Risk matrix (likelihood × impact over a 2–3-year horizon; evidence basis noted; INTERPRETATION built on the FACT record above):

Aggregate read (INTERPRETATION): the catastrophic-loss risk is low — a state-protected oligopoly with 1.6× leverage, AAA domestic credit, and ₩1.6T of breach-year FCF does not go to zero; the realistic bear case is a return to telecom-EPV pricing (~ADR $14–20, i.e., the December 2025 experience, a −35–55% drawdown from $30.62). The dominant risk is not operational; it is paying for the pyramid — risks 1, 2, 5, and 9 are all the same risk wearing different clothes: the AI premium and the chaebol’s claims on the cash machine underneath it.

10. Valuation Discussion

Embedded-expectations analysis: what the current price underwrites, and what must be true. No price target; no recommendation. All math rebuilt from filing-tied figures at ADR $30.62 (2026-07-17), ordinary parity ₩80,469, USDKRW 1,460, 212.98M ordinary shares; market cap ₩17,139B (~$11.7B); EV ₩24,540B (~$16.8B) at Q1-26 net debt ₩7,308.8B + NCI ₩92.2B.

10.1 Own-history multiples: now vs. the December low vs. the June high

A process note first (FACT): the standard AZI own-history valuation-percentile feed is degenerate for SKM — the valuation_index.history array is empty (length 0) and all three percentiles print identically at 96.758 with a stale TTM-EPS denominator ($1.94/ADS ≈ FY2024-level earnings versus true reported TTM ~$0.66/ADS at the breach trough). The “96.8th percentile” figure is a feed fallback, not data; it is not quoted anywhere in this report as evidence. The honest substitute is the rebuilt series below.

What the melt-up was (INTERPRETATION): in five months (Dec 18, 2025 → Jun 2, 2026) the market took normalized P/E from ~9× to ~21×, EV/EBITDA from ~3.5× to ~6.3×, P/B from 0.83× to 1.94×, and the implied dividend yield from 6.4% to 2.7% — a ~2.3× re-rating of the equity with no change in telecom fundamentals (FY2025 was the worst earnings year in company history). The crack gave back roughly half; today’s price sits ~60% of the way from the December-low multiple to the June-peak multiple. This was an AI-option premium cycle, not an earnings cycle. Even after the −33% crack, SKM trades above its entire pre-2026 own-history band (P/B 1.29× vs ~0.8–1.0× through 2022–25; normalized P/E ~14× vs ~9–12×).

10.2 Peer comps (third-party color; prices 2026-07-16/17, TTM Q1-2026 financials, EVs price-scaled off Q1-26 balance sheets; LG U+ corrected for the same “EBITDA = op income” aggregator bug as SKM)

Placement (INTERPRETATION): SKM trades at a clear premium to its Korean oligopoly siblings (KT 8.2×/3.5×, LG U+ 11.5×/3.1× vs SKM 14.2×/4.6×). Historically SKT carried only a slight #1-player premium over KT, so most of today’s gap is the AIDC premium made visible in the comp table. Against the global telecom cluster (~6–8× EV/EBITDA ex-TMUS, ~7–11× P/E, ~4.5–6% yields) SKM still discounts on EV/EBITDA (4.6×) and P/B (1.29×) — but its normalized P/E is now above T and VZ. The “Korea discount” has not disappeared; it has migrated into EV and book terms. TMUS at 20×/10.2× prices durable growth and buybacks SKM does not have; KT and LG U+ are the honest anchor for what a Korean telecom without an AI story is worth.

10.3 Telecom EPV — the regulated pool, valued as what it is

The approach: anchor the earnings-power value on the regulated pool and treat AIDC as optionality (ASSUMPTIONS labeled): normalized operating profit ₩1,700B (FY23 1,756 / FY24 1,691 IFRS), taxed at 23% → NOPAT ₩1,309B. No-growth perpetuity (maintenance capex ≈ D&A in steady state; population shrinking and tariffs capped → g ≈ 0):

Cross-check: normalized net income ₩1,200B × KT’s 8.2× = ₩9.8T equity — dead center of the EPV range. The telecom-only equity is worth ₩9.0–11.3T (₩42,000–53,000 per ordinary share; ADR $16–20). Note what this implies: the December 2025 low ($19.68 ≈ ₩51,719 ordinary parity) was almost exactly telecom-EPV pricing — the market briefly valued SKT as a pure utility and the AI pyramid at zero. That is why the December low is the natural bear anchor rather than an arbitrary “support level.”

10.4 The embedded AIDC premium — reverse-engineering what today’s price pays for

EV today ₩24,540B minus telecom firm EPV ₩16,362–18,700B = ₩5,840–8,177B (~$4.0–5.6B) of AI option value embedded at $30.62 — 34–48% of the equity value. The same calculation at the June high gave ₩14.4–16.8T (~$10–11B); at the December low, ~₩0–2T (≈zero). The premium is the cycle.

What AIDC must deliver to justify ~₩7T of premium (ASSUMPTIONS explicit): compounding the premium at a 10% required return to 2030 implies a terminal AIDC EV of ~₩10.4T. At 10–12× EBITDA and 20–25% EBITDA margins, that requires 2030 AIDC revenue of ₩3.5–5.2T — full delivery of management’s ₩5T-by-2030 AI target at real margins. Against ₩519.9B of 2025 AIDC revenue, that is ~8–10× growth in five years, into a simultaneous build by Naver, KT, LG, Samsung SDS, the state, and the hyperscalers, at a ROIC that must roughly double the telecom core’s 5–6%. And the margin number that would validate the entire calculation is precisely the number management declines to disclose while calling it “already comparable to telecom” (Q1-2026 call). The premium therefore pays upfront for the single least-verifiable figure in the story. Sensitivity on the downside: if AIDC reaches only ~₩2T revenue at ~15% margin by 2030 (₩300B EBITDA, ~₩3T EV then, ~₩2T today), ~₩5T of today’s premium (~₩23,000/ordinary ≈ $9/ADR) has no supporting cash flows.

10.5 Scenario zones (2–3-year view; RANGES, not targets; per ordinary ₩ / ADR $ at USDKRW 1,460)

• BEAR — AI premium evaporates; Korea discount returns: ₩38,000–53,000 / ADR $14–20. Telecom-only EPV (WACC 8–8.5% → equity ₩8.0–9.0T) up to the December-2025-low experience (equity ₩11.0T = EPV at ~7%); normalized NI ₩1.0–1.2T at 7–9× (the KT multiple). Triggers: AIDC margins disclosed and disappointing; a 2027–29 capacity air-pocket; MSIT tariff or 6G-capex rent extraction; KRW to 1,550–1,590 compounds the ADR leg (at 1,588 the zone is $13.4–18.8). Load-bearing assumption: the regulated ₩1.7T operating pool itself holds — the state is the risk to it, not competition.
• BASE — breach recovery completes; telecom normalizes; AIDC contributes modestly: ₩71,000–83,000 / ADR $27–31. EPS restored to ₩5,500–5,800, NI ~₩1.2T at 11–13× (between Korean peers and the global cluster — i.e., a partial AI credit of ~₩2T); ~4–4.5% dividend yield at the zone bottom providing support; Ulsan on track but pre-profit. The current price ($30.62) sits at the TOP of this zone — the market is already paying for clean base-case delivery plus a down payment on the bull case. Load-bearing assumption: the normalized ₩5,500–5,800 EPS is real (Q1-26 annualizes above it — see the open question in Section 10.7).
• BULL — AIDC executes to plan; Value-up re-rating persists: ₩105,000–135,000 / ADR $40–52. By 2028–29 NI ₩1.6–1.8T (telecom ₩1.2T + AIDC NOPAT ₩400–600B) at 14–16× (partial Korea-discount unwind on ≥10% ROE delivery) → equity ₩22.4–28.8T. Requires: AWS/OpenAI MoUs convert to contracted capacity; Ulsan 41MW Nov-2027 on time; AIDC margin disclosure confirming “telecom-comparable”; and SKT not designated the funding vehicle for the 15GW vision — that path converts the bull case into a leverage/dilution bear case. Zone tops around/above the June $46 high. KRW strength toward 1,320–1,350 adds ~8–10% to the ADR leg.

FX dimension for ADR holders (FACT + arithmetic): ADR = ordinary × 5/9 ÷ FX. The won traded 1,322–1,588 over the past 52 weeks — a ±~9% band around 1,460 that moves the ADR ±~9% independent of anything SKT does (at constant ₩80,469: $33.82 at 1,322 vs $28.15 at 1,588). The factor model’s dominant USDollar loading (−0.54) confirms the ADR is materially a KRW proxy. Korean withholding (~22% statutory; 15% + local under the US treaty) and depositary fees further trim the ~4.1% gross dividend run-rate.

10.6 What the market is pricing correctly vs. incorrectly

Priced correctly (INTERPRETATION):

1. The breach was one-time; the franchise is intact. Q1-2026 (op ₩537.6B, +210K net adds, dividend resumed) validated the fast re-pricing off the December low; the market never confused a regulatory punishment with a franchise break. Correct.
2. The core is a regulated utility, and the AI premium is separable from it. Even post-melt-up, EV/EBITDA (4.6× normalized) and P/B (1.29×) keep a Korea/governance discount versus global telecom — the market did not re-rate the telecom pool itself, only layered an option on top.
3. The factor identity. The model files SKM with EM bond/dividend ETFs (USDollar −0.54, beta 0.42, 5-year Sharpe 0.11) — a low-beta EM yield asset. The June tape priced it as an AI-momentum stock; the identity was rented, not owned, and the −33% crack is the concession.

Priced incorrectly / questionably (INTERPRETATION):

1. ₩5.8–8.2T of AIDC option value embedded today (34–48% of equity) against 4% of revenue, undisclosed margins, MoU-stage anchor customers, and a simultaneous sector build landing 2027–29. This is the load-bearing assumption of the bull case, and it prices full delivery of an undisclosed-margin target.
2. The June price treated SKM as an AI-infrastructure stock (21× normalized P/E, 1.94× book) while its factor DNA never changed — narrative multiple and statistical identity diverged maximally at the peak. The residual risk runs both ways: the “AI premium” re-prices on KOSPI theme beta, not on SKT’s own milestones.
3. The governance/related-party discount may be under-priced at the top of the base zone: ₩506.8B Pangyo from SK Inc., ₩1.14T SKHNPS, a dividend suspended within a year of the Value-up pledge, and the 2026 ROE ≥10% pledge tracking a miss (~8%) — minorities are the residual claimant, and the price is not obviously charging for that.
4. The bear side’s load-bearing assumption is that the regulated ₩1.7T pool is a hard floor. It has held for a decade, but the same state that grants the oligopoly caps the rent; a further extraction or a 6G capex recall compresses the EPV itself, not just the premium.

10.7 The open valuation question

Q1-2026 operating profit of ₩537.6B annualizes to ~₩2.15T versus the ₩1.7T normalized base used above. How much of that gap is KT-breach share recapture and appreciation-package rolloff that fades through 2026? If ₩1.9–2.0T is the true post-breach pool, every zone above shifts up ~₩6,000–8,000 per ordinary share (~$2–3/ADR) and the base-zone placement of the current price looks less stretched. The Q2-2026 print (expected early August 2026) is the first clean read.

11. Variant Perception

Consensus belief (INTERPRETATION, from the tape and sell-side color): the 2026 re-rating was an AI-infrastructure story. The narrative runs: Korea is executing a sovereign AI buildout; SKT is its designated telecom champion (MSIT Sovereign AI project, NVIDIA partnerships, OpenAI Stargate MoU, AWS-anchored Ulsan); AIDC revenue growing 35–89% will compound into a ₩5T business by 2030; and the stock deserves to be re-rated from “Korean telecom at 9× earnings” toward “AI infrastructure platform.” The June peak ($46, 21× normalized P/E, 1.94× book) was that belief fully priced; the current $30.62 is the same belief half-priced. Korean and global sell-side commentary through H1-2026 leaned constructive on exactly this framing (third-party color — Asiae, Yonhap, TelecomTV coverage of the January turnaround and the NVIDIA headlines).

Strongest bull case — the pyramid is real and sovereign-mandated. Korea’s AI-computing sovereignty drive is state policy, not a marketing theme; power-rationed Seoul-metro capacity makes entitled data-center capacity genuinely scarce; SK Group holds the binding inputs (SK E&S LNG power, SK Hynix HBM, SK ecoplant construction); AWS anchoring Ulsan is third-party validation with real capital (~$4B); Q1-2026 AIDC revenue +89.3% is what early ramp looks like; and the Q1 run-rate (~₩2.15T annualized operating profit) suggests the post-breach telecom pool is bigger than the normalized base used in this report. On that reading the embedded ₩5.8–8.2T premium is a down payment on a ₩10T+ 2030 AIDC enterprise, the Value-up ROE ≥10% pledge gets delivered by 2027–28, the Korea discount unwinds, and the bull zone ($40–52) is conservative.

Strongest bear case — a utility wearing an AI costume, with a governance leak. The core is a 5–6% ROIC regulated utility with a shrinking customer base and a political price ceiling, worth ₩9.0–11.3T of equity — full stop. Everything above that is narrative: 4% of revenue, undisclosed margins, MoU-stage anchors, commodity GPUaaS, and a simultaneous sector build that lands as 2027–29 supply. Meanwhile the chaebol steadily taxes the minority claim — Pangyo cash to the holdco, ₩1.14T into an SK Hynix affiliate, a dividend suspended within a year of the pledge, a breach-year CEO bonus and promotion — and the 15GW vision dangles a future in which SKT’s balance sheet is volunteered for national-champion duty. On that reading the December 2025 low ($19.68) was not the panic; it was the correct price, and everything above ~$20 is premium in search of evidence.

The factor-positioning read — the identity conflict the tape hasn’t resolved (FACT from the model; INTERPRETATION on it). FactorsToday files SKM as a low-beta KRW/Korea yield proxy: USDollar −0.54 (dominant), Country-Korea +0.32, Market +0.21, DividendYield +0.06, Value/Quality/LowVol/Growth zeroed, beta 0.42, 5-year Sharpe 0.11, ~81% idiosyncratic variance. Its nearest factor neighbors are EM local-currency bond ETFs (LEMB 0.84, EBND 0.84, EMLC 0.83), international dividend/yield ETFs, Iberdrola, and Banco Santander Chile — the model has never seen an AI stock here, even after a +193% AI-narrative run. The June episode was therefore a category error the market briefly made and is now half-correcting: the buyers at $40+ were momentum/theme flows (KOSPI AI complex, ~5× normal ADR volume) renting a yield asset, not owners re-underwriting the business. The regime backdrop is mixed: Value (+1.70 z, 252d) and DividendYield (+1.66 z) are in favor — supportive of SKM’s actual identity — while Communication Services is extremely out of favor (−2.20 z, 63d) and a firming USD (+0.65 z, 21d) is a mild headwind given the −0.54 loading. Short interest (1.35% of float, 1.1 days) is negligible in both directions — no crowded short to squeeze, no capitulated long base either. Net: consensus is offsides not on the business but on the asset class — SKM keeps being traded as Korea/AI beta when its statistical identity is an EM yield instrument with a call option.

The 3–5 assumptions that matter most, with falsifiers:

1. The regulated ~₩1.7T operating pool is the floor. Falsifier (bear): MSIT tariff intervention or a 6G capex recall compressing the pool; two consecutive quarters of operating profit materially below ₩425B ex-one-offs. Falsifier (bull): Q2/Q3-2026 prints confirming a ₩1.9–2.0T+ pool.
2. AIDC margins are real and telecom-comparable. Falsifier (bear): segment disclosure showing colocation/GPUaaS margins materially below telecom, or continued refusal to disclose through FY2026 reporting. Falsifier (bull): disclosed margins ≥ telecom’s ~10% operating margin with contracted-revenue backlog.
3. The anchor-customer MoUs convert to contracted capacity. Falsifier (bear): OpenAI/Stargate lapses or stays MoU through 2027; AWS scope at Ulsan shrinks; KKR talks (follow-up due ~2026-08-10) collapse or price punitively. Falsifier (bull): signed take-or-pay contracts, KKR close on market terms.
4. SKT’s balance sheet is not volunteered for the 15GW vision. Falsifier (bear): any board approval of SKT-funded GW-scale capex beyond the ~₩3.4T committed slice, or a dividend cut explicitly tied to AI capex. Falsifier (bull): 15GW structured with external capital carrying ≥60–70%.
5. The dividend is a real claim again. Falsifier (bear): FY2026 year-end dividend below ₩1,660/ordinary-equivalent pace or another skip. Falsifier (bull): restoration to the ₩3,320–3,540 run-rate plus a countercyclical buyback.

12. Fact vs. Interpretation

13. Open Questions

1. AIDC segment margins and ROIC — the single most load-bearing missing number. Management claims “comparable to telecom” and refuses disclosure (Q1-2026 call). When does a segment appear in the filings (FY2026 20-F?), and what does it show? Everything about the ₩5.8–8.2T premium hangs on this.
2. KKR funding structure (~₩1T rumor; follow-up disclosure due by ~2026-08-10). Price, instrument (equity/JV/project finance), governance rights, and what it validates or reveals about AIDC unit economics. A close on market terms is the bull case’s first external validation; punitive terms or collapse is a bear data point.
3. The true post-breach run-rate. Q1-2026 annualizes to ~₩2.15T operating profit versus the ₩1.7T normalized base. How much is KT-breach share recapture and package-rolloff that fades? Q2-2026 (early August) is the first test; the appreciation-package discounts rolling off through year-end are the swing factor.
4. FY2026 year-end dividend. Q1’s ₩830 resumed the quarterly, but the year-end decision (post-AGM capital-reserve reduction) will show whether the ≥50%-of-adjusted-NI pledge has teeth after the 2025 suspension — and whether the ROE ≥10% Value-up pledge (tracking ~8%) is formally walked back.
5. 15GW funding plan. The June 30 fair disclosure says partner-funded, size/timing TBD. Watch for any structure that routes capital calls through SKT’s balance sheet or SK Broadband — the Pangyo/SKHNPS pattern says the group’s boundary is permeable.
6. Breach litigation tail. Civil suits pending; SKT rejected PIPC mediation (₩300K × 3,998 claimants); the ₩134.8B fine is under administrative appeal (filed January 2026); March-2026 legislation raises repeat-incident surcharges. Quantification is impossible today; a second security incident would be materially worse under the new regime.
7. PSU performance metrics. The 2023+ PSU program’s targets are not disclosed in the 20-F — it cannot be verified whether executive equity pays on ROE/TSR or on AI-narrative vanity metrics. A disclosure request belongs on any engagement list.
8. SK Inc. exchangeable-bond overhang. The holdco has historically monetized its SKT stake via SPV exchangeable structures linked to ADSs (20-F risk factor); current outstanding exposure and any new issuance into the AI-rally strength are unconfirmed.

14. What Must Be True

For the BULL case (ADR $40–52 zone; AIDC executes, Value-up re-rates):

• AIDC reaches ≥₩1.5–2T revenue by 2027–28 with disclosed EBITDA margins of 20–25% (validating “telecom-comparable”), and the ₩5T-by-2030 trajectory stays credible — ~35–45% CAGR from the ₩519.9B 2025 base.
• AWS/OpenAI commitments convert from MoU to contracted, take-or-pay capacity; Ulsan 41MW opens on time (Nov-2027); the KKR (~₩1T) or equivalent external capital closes on market terms, keeping SKT’s own AI capex inside ~₩1.1T/yr and FCF positive through the build.
• The post-breach telecom pool confirms at ₩1.9–2.0T+ operating profit (Q2/Q3-2026 prints), with share ≥42% and ARPU recovering as the appreciation package rolls off.
• The dividend restores to ₩3,320–3,540/ordinary and stays there through the Ulsan peak-capex years; ROE reaches ≥10% by 2027–28.
• Falsification test: if by mid-2027 there is still no AIDC margin disclosure, no contracted OpenAI/AWS capacity beyond Ulsan phase 1, or the dividend is cut/suspended again with AI capex cited — the bull case is dead and the stock is a utility with a marketing department.

For the BEAR case (ADR $14–20 zone; premium evaporates, Korea discount returns):

• AIDC margins, when disclosed, prove to be commodity-colocation/GPUaaS economics (mid-teens or below EBITDA margin) — or remain undisclosed through FY2026 reporting, itself the tell.
• The 2027–29 Korean AI-DC capacity wave (Naver, KT, LG, Samsung SDS, state, hyperscalers) lands into wobbling AI demand — the same hyperscaler-capex air-pocket risk evident in the memory sector’s public capex plans — crushing utilization and pricing just as Ulsan ramps.
• MSIT extracts further rent (tariff pressure, 6G capex recall, security surcharges), compressing the ₩1.7T pool itself.
• SKT is designated the funding vehicle for any GW-scale phase of the 15GW vision — leverage/dilution with the dividend as the shock absorber.
• KRW weakens toward 1,550–1,590, compounding the ADR leg (the zone at 1,588 is $13.4–18.8).
• Falsification test: if AIDC discloses ≥20% EBITDA margins on ≥₩1.5T revenue with a contracted backlog, and the dividend holds ≥₩3,320 through 2027 — the bear case’s “no supporting cash flows” claim fails and the premium is at least partially earned.

For the BASE case (ADR $27–31; the modal path): breach recovery completes to ₩5,500–5,800 EPS; the market grants ~₩2T of AIDC credit (11–13× normalized earnings); Ulsan builds on time but contributes nothing material before 2028; the dividend grinds back toward ₩3,320; and the AI premium continues to inflate and deflate with the KOSPI AI complex rather than with SKT’s own milestones — meaning the entry price, not the thesis, determines the return. Falsification test (both directions): a sustained ₩1.9T+ operating pool shifts the whole frame up; a margin disclosure (either way) collapses the uncertainty the base case is renting.

15. Source Appendix

Full source-level documentation for every claim in this report — primary filings first (FY2025 Form 20-F filed 2026-04-29, the 60-month 6-K stream, Schedule 13D/13G), then earnings materials, transcripts and call summaries, regulatory actions (PIPC/MSIT), industry data, and press, each with URL and access date — is compiled in Appendix B below. Data-quality caveats that a reader must carry into any reuse of this report — the broken ROIC.ai FY2025 EBITDA/FCF-yield fields, the per-ADS basis of ROIC per-share data, the degenerate azitrading.com valuation-percentile series, and the K-IFRS/IFRS operating-profit presentation gap — are documented in Section 6.5 above and cross-referenced in the appendix.

The analytical body (Sections 1–15) carries no recommendation and no price target; the only opinion expressed anywhere in this report is the labeled Claude’s Take block at the front, which is the author’s own independent view.

APPENDIX A — Standard Diligence Questionnaire

SK Telecom Co., Ltd. (NYSE: SKM) — Standard Diligence Questionnaire

Independent fundamental research · Report date: 2026-07-19 · Coverage type: INITIATION (no prior SKM report)

This appendix answers the Standard Diligence Questionnaire, grounded in the analysis above.

Label key: (F) Fact — tied to a primary or documented source; (I) Interpretation — the analyst’s judgment built on facts; (A) Assumption — a working input used for scenario/valuation math, not established.

No BUY/SELL recommendation and no price target appear in this appendix — no exception. Where valuation is discussed, it is framed as embedded expectations and scenario zones (ranges, not targets).

Basis note (read first — it matters throughout): All financial statement figures are KRW (₩), consolidated, FYE December 31, K-IFRS with the 20-F reconciliation to IFRS-IASB. Per-share figures are stated per ordinary share (KRX 017670) unless marked per-ADS. 1 ADS (NYSE: SKM) = 5/9 of one ordinary share (FY2025 20-F cover); equivalently, 1 ordinary share = 1.8 ADS. ADR prices are USD per ADS. FX context: USDKRW ~1,443–1,488 in mid-2026; 1,460 used where a single FX input is needed (A). Ordinary shares outstanding: 212,982,275 (2025-12-31). Data-quality caveat: ROIC.ai’s FY2025 EBITDA, net-debt/EBITDA, EV/EBITDA, and FCF-yield fields for SKM are broken (D&A missing / sign errors) and were not used; all such figures here are rebuilt from 20-F/EDGAR data.

General

What thoughtful questions have other investors asked about this company?

The recurring institutional questions about SKM in 2026, each with where the evidence stands:

1. Is the ₩5T AI-revenue-by-2030 target real? (F) AIDC revenue was ₩519.9B in FY2025 (+34.9%) and ₩131.4B in Q1-2026 (+89.3%); AI in total (AIDC + AIX) is ~4% of consolidated revenue. (F) Milestones are real: Ulsan AIDC JV with AWS (103MW initial, construction started Sep-2025, 41MW by Nov-2027, full by Feb-2029, >1GW ambition), OpenAI “Stargate Korea” LoI/MoU (Oct-2025), Haein GPU cluster, NVIDIA “AI factory” cooperation (2027, clarified as not GW-scale at first), and a 15GW national buildout vision announced Jun-30/Jul-2026 with Phase 1 (5GW) from 2029. (I) The ₩5T/2030 target requires roughly 8–10x growth in five years into a simultaneous build by Naver, KT, LG, Samsung SDS, the Korean state, and hyperscalers — a textbook capital-inflow wave. The industry read is ~60% real economics (state-backed demand, AWS anchor, SK Group inputs) / 40% chaebol re-rating narrative (MoU-stage anchors, no committed capital behind OpenAI, group synergies accruing to the chaebol rather than SKM minorities). The target is a stretch goal, not a plan with contracted demand behind it.

2. What are AIDC margins? (F) Undisclosed — AIDC is not a reported segment, and management (Q1-2026 call) claimed AIDC profitability is “already comparable to telecom” while refusing to disclose profitability metrics. (I) This is the single most load-bearing missing number in the story: the market is paying an AI premium upfront for a margin claim the company will not show. Colocation/GPUaaS on a small, high-utilization base can print good margins; hyperscale economics at 100MW+ with debt are different.

3. Is the 2025 breach truly one-time? (F) FY2025: revenue ₩17,099.2B (−4.7%), IFRS operating profit ₩1,048.5B (−38.0%), net income to parent ₩408.4B (−67.3%); components were the ₩134.8B PIPC fine, a ~₩500B Customer Appreciation Package, free USIM replacements for ~23M subscribers, a 7-week new-subscription suspension, and ~0.73M subs lost in the Apr 19–Jul 14 window. (F) Q1-2026: operating profit ₩537.6B (+351% YoY), +210K handset net adds, dividend resumed at ₩830/quarter. (I) Predominantly one-time: the costs were taken upfront in FY2025 and the competitive damage was small and partially reversed (see moat section). The residuals are a litigation tail (mediation proposals rejected; civil suits pending), ~₩700B of cybersecurity investment over 5 years (mostly 2026+), and permanently higher security opex — an earnings drag, not a recurring crisis.

4. How much of the embedded AI premium is justified? (F) Reverse-engineered from the Q1-2026 balance sheet: EV at the $30.62 ADR is ~₩24.5T vs. a regulated-telecom EPV of ₩16.4–18.7T, leaving ~₩5.8–8.2T (roughly $4.0–5.6B, 34–48% of equity value) of AIDC optionality embedded today (I, with A inputs: WACC 7–8%, normalized op profit ₩1.7T). At the June high it was ₩14.4–16.8T; at the Dec-2025 low it was ~₩0–2T. (I) Justifying ~₩7T requires 2030 AIDC revenue of ₩3.5–5.2T at 10–12x EBITDA and 20–25% margins — i.e., full delivery of management’s target at real margins, against ₩520B of 2025 revenue and undisclosed profitability. If AIDC reaches only ~₩2T revenue at ~15% margin, ~₩5T of today’s premium has no supporting cash flows.

5. What does the chaebol structure cost minorities? (F) The record: the 2021 SK Square spin-off moved the 20.1% SK Hynix stake (SK Square NAV now ~₩273T, itself trading at a ~41% holdco discount) out of SKT; SK Broadband bought the Pangyo data center from SK Inc. for ₩506.8B (Jun-2025, common-control transfer at carrying amounts); SKT committed ~₩1.14T into SK hynix affiliate SKHNPS/Solidigm (Jun-2026); SK Inc. brand/IT fees ₩440.2B (2025); SK Inc.'s stake crept from 26.8% to 30.6% via SKT’s own treasury retirements at zero cost to the holdco. (I) The cost is real but mostly opportunity cost and governance discount rather than outright theft: the spin-off was pro-rata in form (a 2021 holder who kept both legs owns the Hynix windfall), but SKT minorities were left holding the entity stripped of the group’s best asset and its ~₩1T/yr equity-method income, and cash flows up the chain (Pangyo) when the group wants it. The hierarchy of claimants is: national AI champion > SK Group > minorities.

6. Will the dividend fully restore to ₩3,540? (F) FY2023/2024 DPS was ₩3,540; FY2025 was ₩1,660 (Q3-2025 interim and year-end both suspended); Q1-2026 resumed at ₩830/quarter (₩3,320 annualized run-rate). Policy is ≥50% of adjusted consolidated net income for 2024–2026; management says it “aims to restore payout levels over time.” Mar-2026 AGM approved a capital-reserve reduction (return capacity). (I) Full restoration to ₩3,540 is plausible on normalized earnings (₩3,540 was 58% of FY2024 NI; the resumed ₩830/qtr is already 94% of the way there), but the breach year proved the dividend is the swing variable when group priorities collide with it — expect restoration to be contingent on the AIDC build not demanding the cash first.

7. Is SKM a utility with an option, or an AI stock with a utility attached? (F) The factor model files SKM with EM local-currency bond ETFs and international dividend/yield ETFs (cosine similarity ~0.83–0.84), not with AI/tech names: USDollar loading −0.54, Korea +0.32, DividendYield +0.06, beta 0.42, 5-year Sharpe 0.11. 96% of revenue and essentially all earnings are the regulated telecom core. (I) Economically it is a regulated utility with an AI option; the June-2026 tape briefly priced it as the reverse, and the −33% crack was the market conceding the point. The variant perception is precisely this identity gap.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? (F) Earnings are at an idiosyncratic trough — FY2025 operating profit −38% and net income −67% on the breach — with recovery underway: Q1-2026 operating profit ₩537.6B (+351% YoY, −5.3% vs Q1-2025), Q1 net income ₩322.4B. This is not a macro-cycle trough: 2025 was a company-specific punishment year, and in fact the KOSPI rose +75.6% in 2025 while KRX SKT fell −3.1%. (I) The right framing is not “where in the cycle” but “what is the true post-breach run-rate”: the normalized base is ~₩1.7T operating profit (FY23 ₩1,756B / FY24 ₩1,691B IFRS), while Q1-2026 annualizes to ~₩2.15T. (A/OQ) How much of the gap is KT-breach share recapture that fades is unresolved; if ₩1.9–2.0T is the real post-breach pool, every valuation zone shifts up ~₩6,000–8,000 per ordinary share.

Driven by the external environment or the company’s own actions? (F) Both, but the distinguishing driver was company-specific: the breach (a security failure — 28 of 42,605 servers infected, plaintext credentials, per the MSIT probe) and the company’s own remediation choices drove FY2025; external regulation (PIPC fine, MSIT-ordered ETF waivers) set the punishment. The Q1-2026 recovery is partly external — KT’s own breach (Sep-2025) handed SKT two-thirds of KT’s ETF-waiver outflow in Jan-2026. (I) The underlying telecom earnings pool is remarkably insensitive to the macro environment; what moves it is the state’s rent extraction and company-specific execution, not GDP.

How stable are revenues? (F) Extremely stable: consolidated revenue has been in a ₩16.7–17.9T band for five straight years (FY2021 16,748.6 → FY2024 17,940.6 → FY2025 17,099.2), and OCF has held ₩3.9–5.8T every year 2016–2025. Demand is non-discretionary and tariff-regulated; even Korea’s largest data breach moved revenue only −4.7%. (I) This is utility-grade stability — but note it comes with zero operating leverage: FY2025 revenue is barely above restated FY2019, and margins are capped ~9–10% by MSIT intervention, so stability means permanence, not compounding.

Outlook for products/services? (F) Wireless (59% of revenue, K-IFRS segment basis): 5G migration is mature (33.85M subs = 65.4% of population by May-2024; nationwide coverage Apr-2024); ARPU fell 5.1% to ₩27,845 in FY2025 on the appreciation package; the game is mix/ARPU recovery, not volume. Fixed-line/broadband/media (SK Broadband, now wholly-owned): flat at ~₩2.5T broadband+media. The only growth pocket is fixed-line misc/B2B/infra (+12.4% to ₩1,327.2B — where data-center revenue lives) and AIDC (+34.9%/+89.3%). (I) Core services: flat forever, by design. All forward growth is the AIDC bet.

How big will this market be — growing, shrinking, domestic or international? (F) Korean telecom + pay-TV service revenue is forecast at ~1.9% CAGR 2024–2029 (GlobalData); mobile data ~5.6% CAGR; the market is almost entirely domestic (SKM has no meaningful international operations). Korea’s population peaked in 2020 and is declining with the world’s lowest fertility rate; the subscriber base is >100% penetrated. (I) A slowly shrinking, saturated, domestic market for the core — the AIDC pivot is management’s answer to that arithmetic, and it is also domestic (Korean AI infrastructure), so the population/scale constraint reappears at the group level.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? (F) Korean mobile is a closed 3-license oligopoly: SKT ~47% / KT ~32% / LG U+ ~21% (2025 est.), with a weak state-supported MVNO tail, no cable bypass, and no fourth entrant ever licensed — spectrum is assigned by MSIT. The 2014 Handset Reform Act capped device subsidies, ending subsidy wars. (I) Structurally less competitive over time on price (the state suppressed the one competitive lever), but the arena is shifting: competition is migrating to AI infrastructure, where every chaebol, Naver, and the state are building simultaneously — so “less competitive telecom, more competitive adjacency.”

How profitable is the business (ROIC, ROE)? (F) ROIC 5.4–6.1% (2022–2024), 2.2% in the FY2025 breach year; ROE 5.45% FY2024, 1.78% FY2025; operating margin ~9–10% steady 2016–2024. (I) This fails Greenwald’s franchise test (15–25% sustained ROIC = advantages present; 6–8% = absent). The moat confers stability, not super-normal returns, because the same state that blocks entry caps the rent (SKT, as dominant carrier, needs MSIT tariff approval). Utility-level economics, and below any reasonable WACC.

How profitable is the industry — how many competitors, what barriers to entry? (F) Three carriers in stable 47/32/21 proportions; SKT #1 for ~40 years with <2pp share movement over any 5–8 year window (passes the share-stability test). Entry barriers are near-absolute: three licenses, state-assigned spectrum, ₩-trillion fixed-network costs. (I) Industry profitability is mediocre because of the barrier’s source: regulation-conferred oligopoly economics — the state is the fourth player and the marginal claimant on rent (tariff approval, subsidy caps, fines, license obligations, exclusion of MNOs from the private-5G market). Returns are worse than the US three-player market (per prior analysis of the US carriers’ public filings) because the Korean state captures more of the oligopoly rent for consumers.

Can the business be easily understood? (F/I) Yes for the core: a domestic wireless/fixed-line utility with a 40-year #1 share — subscriptions in, network costs out. The complications are (1) chaebol governance/related-party flows and the 2021 spin-off history, (2) the undisclosed economics of the AIDC pivot, and (3) K-IFRS vs IFRS presentation gaps (the ₩134.8B fine sits below the domestic K-IFRS operating line, so SKT’s headline op profit excludes it). An investor modeling only the K-IFRS headline will misstate breach-year economics.

Can it be undermined by foreign low-cost labor? (F/I) No — this question maps poorly to a domestic regulated service utility. The correct analog is technology/bypass risk (satellite, private 5G, OTT substitution for messaging/voice) and regulatory bypass (state-sponsored MVNOs). None is currently material: the MVNO tail is weak, private-5G is fenced off from MNOs by the state, and network access is non-substitutable. AIDC/GPUaaS, however, is exposed to global cost curves — GPU rental is commodity compute with no captivity, competing against hyperscaler economics worldwide.

Do brands matter? (F/I) The 2025 breach was a natural experiment: Korea’s largest data breach, a record fine, free replacements for 23M customers — and SKT lost only ~0.8M subscribers (~3% of base) in a state-forced open window, then recaptured share from KT’s own incident. (I) Brand proved non-load-bearing: what retains customers is the closed three-license loop, bundle/habit captivity, and the fact that rivals carry identical risk. The brand was damaged and it didn’t matter competitively.

What is the nature of competition? (F) Infrastructure scale and regulatory position, not price: no price wars currently (state-capped subsidies; SKT tariffs need MSIT approval); competition expresses itself in network quality, bundles (mobile + B tv IPTV + broadband), and marketing intensity. (I) Oligopolistic coexistence under state supervision — the carriers even coordinate visibly when it suits the state (simultaneous 5G launch, 2019).

Customers’ switching costs? (F) Low in theory — number portability since 2004 — and the state can and did lower the drawbridge (MSIT ordered ETF waivers post-breach). What remains: family/combined bundle discounts, membership habit, inertia. (F) But the breach proved positional captivity: churned subscribers rotate among the same three carriers (closed-loop churn) — when KT waived ETFs in Dec-2025 it lost 216K subs faster than SKT did, and two-thirds of the Jan-10 outflow ported to SKT. (I) Switching costs are shallow at the individual level and near-absolute at the system level: you can leave SKT, but you cannot leave the oligopoly.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? (F/I) (1) The AIDC optionality — entitled/powered land positions, the Ulsan JV, GPUaaS infrastructure, and stakes (Penguin Solutions $200M, Lambda, Perplexity) — carries value not visible in book terms; LT investments of ₩5,427.4B are carried at cost/equity-method, not market. (2) SK Group synergies (LNG power adjacency via SK E&S, SK Hynix HBM supply, SK ecoplant construction) are real inputs to the AIDC thesis that appear nowhere on SKM’s sheet — and, note, accrue partly to affiliates, not to SKM. (3) Spectrum/frequency rights are on the sheet but at cost (₩664.5B). (I) None of this is hidden value in the classic sense — it is option value whose realization depends on execution and on how much the chaebol lets minorities keep.

Off-balance-sheet liabilities? (F) Modest and conventional: lease liabilities ₩1,525.8B are on the balance sheet (IFRS 16); the ₩398.5B hybrid bonds (2023) are equity-classified but pay ₩19.8B/yr and are economically debt-like. Contingent items: breach civil litigation (unquantified; mediation proposals of ₩300K × 3,998 claimants and ₩100K × 58 rejected), the ₩738.4B SKHNPS capital commitment deferred to Jun-2030, and the uncosted 15GW AIDC ambition. (I) The real “off-balance-sheet liability” is the unquantified future claim of the AIDC program on SKT’s balance sheet — partner-funded per disclosure, but the Pangyo precedent shows group priorities can re-route SKT cash.

How conservative is the accounting? (F) Fairly conservative on the breach: costs were taken upfront in FY2025 (fine expensed and provisionally paid; ~₩500B package absorbed in H2-2025 revenue), and the fine is non-deductible, producing a 48.1% FY2025 effective tax rate vs 21–23% normal — no smoothing. (F) Caveats: the fine sits below the domestic K-IFRS operating line (headline-flattering); management’s “AIDC profitability comparable to telecom” is an unverifiable claim without segment disclosure; PSU performance metrics are not disclosed in the 20-F. (I) Statement quality itself is high — revenue/NI/OCF/capex/dividends tie exactly to EDGAR XBRL for FY2020–FY2025 — but presentation choices and disclosure gaps favor the company at the margins.

How CapEx-hungry is the business? (F) Core telecom capex is falling: ₩3,687.8B (2020) → ₩2,973.9B (FY23, 20-F PP&E basis) → ₩2,487.4B (FY24) → ₩2,206.6B (FY25), with wireless-network capex nearly halved to ₩733.9B as the 5G build finished. (F) But SKT’s committed AIDC slice is ~₩3.4T through 2028 (~₩1.1T/yr), and FY2026 guidance contemplates “additional AI-related investments.” (I) The business is transitioning from declining capital intensity to a new capital call: FCF (₩1.6–2.5T/yr) can absorb the committed program alongside an ₩800B/yr dividend if AWS/partners carry ~60%+ of Ulsan and telecom EBITDA holds — but the 15GW vision (~$10B+/GW industry cost) is categorically not fundable by SKT and implies future leverage, JV/project-finance, or dilution. Watch the FCF inflection as Ulsan construction peaks into 2027 operations.

Balance-sheet strength (the question behind the questions): (F) Net debt ₩7,357B at YE2025 (₩7,309B Q1-2026) ≈ 1.6x corrected EBITDA (true FY2025 EBITDA ~₩4,604B = IFRS op 1,048.5 + D&A 3,555.6; normalized ~₩4.6–5.3T — note ROIC’s 7.25x figure is a broken feed). Gross debt ₩10,372.8B (debentures 8,213.9 + borrowings 633.1 + leases 1,525.8); 16.2% USD-denominated; interest expense ₩383B vs EBITDA/interest ~12x. Ratings: Moody’s A3 / S&P A− / Fitch A−, domestic AAA. (I) Solid investment-grade utility balance sheet with ample capacity for the committed AIDC build; leverage is not a near-term risk — funding structure for the uncommitted vision is.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy? (F) FCF ₩1.6–2.5T/yr (FY2021 1,722.8 / FY2022 2,112.9 / FY2023 1,866.6 / FY2024 2,528.1 / FY2025 1,600.6 — positive even in the breach year). Uses: dividends (₩628–804B/yr cash paid), occasional buybacks, the SK Broadband minority buyout (₩1,146B, 2024–25), related-party deals (Pangyo ₩506.8B; SKHNPS ₩1.14T committed), AI stakes, and now the AIDC build. (I) Philosophy: competent stewardship of the cash machine at the operating level (capex discipline ₩3.7T→2.2T, deleveraging to ~1.3–1.6x), but the cash machine belongs to the chaebol first — dividends are paid when convenient and were the swing variable the moment breach costs and AI capex arrived.

Significant acquisitions recently? (F) Yes, and the mix is telling: (1) SK Broadband minorities: ₩1,145.9B for 24.76% (Nov-2024, paid May-2025), then the remaining 0.9% cashed out at ₩15,032/sh (completed 2026-05-29) → wholly-owned — consolidates the DC vehicle ahead of the AIDC push; (2) Pangyo Data Center from SK Inc.: ₩506.8B (Jun-2025) — a common-control transfer, i.e., cash from SKT minorities’ consolidated entity up to the holdco; (3) SKHNPS/Solidigm: ~₩1.14T committed (Jun-2026) into a US SK hynix affiliate “to facilitate synergies with AI business” — ₩738.4B deferred to 2030; (4) AI stakes: Penguin Solutions $200M convertible preferred (Dec-2024, 10.3% as-converted, one board seat), Lambda $20M, Perplexity $10M. (I) The two big related-party items (Pangyo, SKHNPS) look more like chaebol capital recycling than arm’s-length investment; their pricing fairness is the outstanding minority-protection question.

Buying back shares? (F) Episodically and opportunistically: ₩426.7B (2020), ₩285.5B in 2023 at ~₩49.4k avg via an SK Securities (affiliate) trust — well-timed ex-post vs ~₩80k today — plus a ₩2.6T treasury retirement (2021) and 4.04M-share cancellation (2024). Zero buybacks in 2025 and none announced 2026 YTD — the program stopped entirely when the stock was cheapest (ADR 52-week low $19.68 in Dec-2025). (I) Buybacks are flow-dependent, not countercyclical; note also that treasury retirements mechanically raised SK Inc.'s stake (26.8%→30.6%) at no cost to the holdco.

Issuing large amounts of new shares to insiders? (F) No — no Section 16 stream exists (FPI), no dual class, directors and officers hold <0.1% (172,012 shares in aggregate). Treasury-settled PSU grants are modest (~0.4–0.5M sh/yr, ~0.2% of shares). (I) This is one place minorities are not diluted — control economics run through the SK Inc. stake, not through SKT-level insider grants.

Compensation policy of directors/management? (F) Thin FPI disclosure: aggregate director comp FY2025 ₩5.3B; executive officers ₩60.4B. Outgoing CEO Ryu Young-sang received ₩3,407M including a ₩1.8B bonus in the breach year (NI −73%, Korea’s largest data breach) — and was moved to the group Supex Council, not fired. New CEO Jung Jai-hun ₩2,079M (from Oct-2025). PSU program since 2023 (3-yr vesting, treasury-settled; CEO eligible up to +100% of salary) — performance metrics are not disclosed in the 20-F (cannot verify ROE/TSR linkage; disclosure gap). (I) Compensation is modest by US standards but accountability-free by US standards too: the breach-year bonus and soft landing say the incentive system answers to the group, not to shareholders.

Motivations of management? (F/I) The observable hierarchy: national-champion status and SK Group strategy first (the AI pyramid, the 15GW vision, related-party support of SK hynix’s orbit), minorities second (dividends restored when affordable, suspended when not). The post-breach CEO swap to Jung Jai-hun — an ex-judge and former Chief Governance Officer, effective at the 2026-03-26 AGM (announced in the Oct-30-2025 reshuffle) — reads as regulatory/reputational repair. The Value-up pledge (≥50% of adjusted NI returned 2024–26; ROE >10% by 2026; ₩30T revenue/35% AI by 2030) is genuine Korea value-up-script participation — but FY2026 ROE is tracking ~8%, a likely miss, and the payout pledge was breached within a year of being made.

Dividend record (the binding promise): (F) ₩3,540 FY23 → ₩3,540 FY24 → ₩1,660 FY25 (Q3 interim and year-end suspended — the first suspension in the quarterly-dividend era) → ₩830/qtr resumed Q1-2026. FY2025 cash dividends of ₩628B were 168% of breach-year NI — the cut still out-earned the P&L. (I) Restoration toward ₩3,320–3,540 annualized is underway; treat the policy as “≥50% of adjusted NI subject to group capital needs,” because that is how it behaved under stress.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? (F) ADR. SKM is the NYSE-listed ADS of a Korean foreign private issuer (20-F/6-K filer, CIK 0001015650), US-dollar denominated; 1 ADS = 5/9 of one ordinary share (par ₩100) listed as KRX 017670. No K-1, no partnership taxation. Consequences for US holders: Korean dividend withholding (~22% statutory incl. local sutax; ~15% + local under the US treaty), depositary fees, full KRW translation exposure on price and dividends (KRW traded 1,322–1,588 over 52 weeks ≈ ±9% ADR band independent of fundamentals), and US-hours gapping on overnight KRX/FX moves. The KRX line (~₩200B+/day) is the deeper venue than the ADR (~$74M/day). ADRs were delisted from the LSE in Jul-2023.

Dividend policy? (F) ≥50% of adjusted consolidated net income returned to shareholders (dividends and/or buyback-cancellation), 2024–2026, per the Oct-2024 Corporate Value-Up Plan; quarterly cadence since 2021. Current run-rate: ₩830/quarter ordinary = ₩3,320/yr ordinary ≈ ₩1,844/ADS ≈ $1.24/ADS ≈ 4.1% gross yield at $30.62 (before withholding and depositary fees). (F) The policy was suspended in practice for two quarters of FY2025. (I) Credible as a floor on normalized earnings; conditional under group capital stress.

How profitable is the business? (F) On normalized (ex-breach) basis: operating profit ~₩1.7T, net income ~₩1.2T, EPS ₩5,500–5,800/ordinary (A — the ex-breach normalized base); op margin ~9–10%; ROIC 5–6%; ROE ~8–9%. Multiples at $30.62 (2026-07-17; normalized basis, FX 1,460): P/E ~13.9–14.6x, EV/EBITDA ~4.6x (5.3x on FY2025), P/B 1.29x, FCF yield 12.7% normalized / 9.3% FY2025, dividend yield ~4.1% restored run-rate. Reported-TTM P/E is ~47x — a meaningless trough-E artifact. (F) Peer placement: KT 8.2x/3.5x/4.6% yield; LG U+ 11.5x/3.1x/4.5%; T 7.1x/6.9x/5.2%; VZ 10.7x/7.8x/6.2%; TMUS 20.1x/10.2x/1.9% (P/E / EV-EBITDA / yield). (I) SKM trades at a clear premium to its Korean siblings — historically only a slight #1-player premium — so most of that gap is the AIDC premium made visible; vs the global cluster the Korea discount now lives in EV/EBITDA and P/B, not the earnings multiple. Data caveat (F): AZI’s own-history valuation percentiles are degenerate for SKM (empty history array; identical 96.758×3 print — a feed fallback, not a real percentile; do not quote “96.8th percentile”). Rebuilt honestly, SKM still trades above its 2022–2025 own-history band (P/B 1.29x vs 0.8–1.0x; normalized P/E ~14x vs ~9–12x) even after the −33% crack.

Is net income diverging from cash from operations? (F) Yes, structurally and benignly: FY2025 NI to parent ₩408.4B vs OCF ₩3,923.8B — a ~9.6x gap. The driver is D&A of ~₩3.6T/yr against a ₩11.9T net PP&E base — the classic D&A-heavy telecom pattern — plus FY2025-specifics: the non-cash portion of breach provisioning, and the high ETR (48.1%) crushing reported NI without touching cash. (F) FCF stayed positive at ₩1,600.6B even at the trough; OCF has held ₩3.9–5.8T for a decade. (I) The divergence is the good kind — earnings understate cash generation — with two caveats: (1) capex absorbs ₩2.2–3.1T/yr of it, so FCF (₩1.6–2.5T) is the honest number, and (2) if AIDC capex ramps, D&A will lag the cash outflow and reported earnings will overstate near-term FCF — the divergence will invert in appearance during the build.

Risks & Downside

What factors would cause the stock to decline?

1. AIDC execution/funding risk (F/I): the premium is ₩6–8T of embedded optionality against 4% of revenue, undisclosed margins, and MoU-stage anchors. A capacity wave lands 2027–29 (Naver, KT, LG, Samsung SDS, state, hyperscalers all building simultaneously); a utilization/pricing air-pocket — or SKT being designated the group’s funding vehicle for the 15GW tail — converts the bull case into a leverage/dilution bear case. KKR ~₩1T funding rumor follow-up due ~2026-08-10 (6-K 2026-07-10) is the next disclosure checkpoint.
2. Regulation extracting more of the ₩1.7T pool (F/I): the state grants the oligopoly and caps the rent — tariff approval, subsidy caps, fines, a possible 6G capex recall (MSIT wants 6G by 2029; carriers are resisting), and new Mar-2026 laws raising CISO accountability and repeat-incident surcharges. This attacks the EPV floor itself, not just the premium.
3. FX (F): KRW traded 1,322–1,588 over 52 weeks — a ±~9% ADR move independent of fundamentals (USDollar loading −0.54). A slide to 1,550–1,590 compounds any KRX decline for ADR holders.
4. Chaebol related-party leakage (F/I): Pangyo ₩506.8B, SKHNPS ₩1.14T, ₩440B/yr brand/IT fees to SK Inc.; plus SK Inc. exchangeable-bond monetization of its 30.6% stake as an ADS-overhang risk (20-F risk factor). Each leak is small vs market cap; the pattern is the discount.
5. Breach-recovery shortfall (I): Q1-2026 op profit annualizes ~₩2.15T vs the ₩1.7T normalized base; if the KT-breach recapture fades, or litigation tail costs exceed expectations, the recovery narrative re-rates down. Civil suits remain pending.
6. Population decline / saturated domestic demand (F): volume growth is structurally over; everything depends on ARPU under a political price ceiling.
7. KOSPI theme-beta (F): the June crack (−9.99% KOSPI circuit breaker 2026-06-23; SKM −33.4% off the $46 high) showed SKM now trades partly as Korea/AI beta — it can fall on theme unwinds with no company news at all.

Risk of a catastrophic loss? (I) Low. The core is a state-protected three-license oligopoly with non-discretionary demand, ~1.6x net leverage, OCF of ₩3.9–5.2T/yr, and FCF positive even in the worst year in company history. The state punishes SKT (fines, ETF waivers) but also needs it (national champion, 6G, sovereign AI). Bankruptcy scenarios require imagination beyond the evidence.

Chance of a total loss? (I) Effectively nil for the reasons above. The realistic bear case is not zero — it is multiple/premium compression: telecom-only EPV of ₩8.0–11.3T equity (₩38,000–53,000/ordinary, ADR ~$14–20), i.e., roughly −35% to −55% from $30.62 as the AI premium evaporates toward the Dec-2025 zero-premium anchor, possibly compounded by KRW weakness. That is a large drawdown, but it is a valuation outcome, not a solvency one. (A — scenario zone, not a target; inputs: WACC 8–8.5%, normalized NI ₩1.0–1.2T at 7–9x, KT-anchored.)

Recent News & Events

Has the business environment changed recently? (F) Yes — three structural shifts in 18 months: (1) the breach regime: Apr-2025 incident → guidance cut Jul-2025 → record ₩134.8B PIPC fine Aug-2025 → mediation rejected Nov-2025 → admin lawsuit filed Jan-2026; industry-wide tightening followed (KT’s own breach Sep-2025; new accountability laws Mar-2026). (2) The AI-infrastructure pivot became the corporate center of gravity (AI reorg Sep-2025; ₩5T/5-yr investment; “AI company-in-company”). (3) The market environment: Korea’s value-up trade + KOSPI AI melt-up (+75.6% in 2025, then a −9.99% circuit-breaker crack 2026-06-23) re-rated and then un-re-rated SKM’s AI premium.

Significant acquisitions? (F) SK Broadband minority buyout completed 2026-05-29 (₩1.146T total for 24.8%+0.9% → wholly-owned); Pangyo DC from SK Inc. ₩506.8B (2025); ~₩1.14T SKHNPS/Solidigm commitment (Jun-2026, share purchase Jul-2026, capital contribution to 2030); Penguin Solutions $200M (Dec-2024). Divestitures: SK stoa/Media S commerce subs (2025), NATE/SK m&service, Kakao shares (2025), Joby shares. (I) Portfolio is being pruned of commerce/legacy and concentrated on telecom + AI infrastructure — rational in direction, chaebol-flavored in execution.

Change in accounting policies? (F) None material identified in the corpus. Note (not a policy change): the K-IFRS-vs-IFRS presentation gap means the ₩134.8B fine sits below the domestic headline operating line; and the Mar-2026 AGM’s capital-reserve reduction is a capital-mechanics change enabling future returns, not an accounting-policy change.

Recent changes — new markets, facilities, management? (F)

• Management: CEO Ryu Young-sang → Jung Jai-hun (reshuffle announced 2025-10-30; representative director confirmed 2026-03-26) — a legal/governance profile, post-breach repair; CFO Kim Yang-seob → Park Jong-seok.
• Facilities/markets: Ulsan AIDC construction start Sep-2025 (AWS JV; ops from 2027); Haein GPU cluster (>1,000 B200s, Aug-2025, MSIT Sovereign AI project); OpenAI MoU Oct-2025 (two facilities from 20MW, southwest Korea); Samsung 6G/AI-RAN MoU Nov-2025; NVIDIA “AI factory” cooperation (first factory 2027, clarified not GW-scale initially, 6-K 2026-06-08); 15GW AIDC vision (fair disclosure 2026-06-30: 5GW from 2029, +10GW from 2035, partner-funded, size TBD); KKR ~₩1T investment rumor (6-K 2026-07-10: nothing decided; follow-up by ~2026-08-10).
• Ownership tape: Wellington crossed 5% (13G 2026-05-15, 5.4% passive); NPS trimmed to 6.7% (from 8.8%); foreign ownership ~36.2% vs 49% statutory cap.
• Results/returns: Q1-2026 print (2026-05-07): revenue ₩4,392B (−1.4%), op profit ₩537.6B (+351% YoY vs trough quarter; −5.3% vs Q1-25), NI ₩322.4B; dividend resumed ₩830/qtr (2026-04-27).
• Price action: +193% melt-up off the Oct-2022 low to $46.00 (2026-06-02) on ~5x participation, then −33.4% to $30.62 (2026-07-17) on the KOSPI AI crack; YTD still +49.1%; short interest ~1.35% of float (no squeeze dynamics).

All figures are from the public sources listed in Appendix B (20-F/6-K primary where available); broken third-party aggregator fields were excluded and rebuilt by hand. No recommendation; no price target.

APPENDIX B — Source Appendix

SK Telecom Co., Ltd. (NYSE: SKM) — Source Appendix

All sources accessed 2026-07-19 unless noted. Primary listing KRX: 017670; ADR NYSE: SKM; CIK 0001015650; CUSIP 78440P306. Foreign private issuer (FPI): files Form 20-F annual reports and Form 6-K current reports only. FYE Dec 31; KRW; K-IFRS (20-F reconciled to IFRS as issued by the IASB).

1. Primary — SEC filings (CIK 0001015650)

EDGAR company page: https://www.sec.gov/edgar/browse/?CIK=1015650

Corpus reviewed: 195 documents (180 Form 6-K + 9 Form 6-K/A + 5 Form 20-F + 1 SD; 13D/13G/Form 3 index entries excluded) retrieved from SEC EDGAR on 2026-07-19. Caveat: only each filing’s primary document was reviewed; ~50 pre-2024 6-K covers are boilerplate-only and their substance sits in exhibits (the 20-Fs cover those events retroactively — e.g., the 2021 spin-off).

1.1 Form 20-F annual reports (all five in the 60-month window)

The FY2025 20-F is the source of record for: ADR ratio (1 ADS = 5/9 ordinary share, cover/Item 12.D — verified), share counts (214,790,053 issued / 212,982,275 outstanding / 1,807,778 treasury at 2025-12-31), ownership (SK Inc. 30.6%, NPS 6.7%), board/audit committee, related-party transactions (Item 7.B), breach-cost detail (Items 3.D/4.B/5.A), K-IFRS↔IFRS operating-profit reconciliation (p.55), dividend table (Item 8.A), capex (Item 5.B), and litigation tail (Item 8.A).

1.2 Latest earnings 6-K

• Q1-2026 preliminary results, filed 2026-05-07, accession 0001193125-26-210347: revenue ₩4.392T (-1.4% YoY), OP ₩537.6B (-5.3%), net income controlling ₩322.4B (-11.5%). https://www.sec.gov/Archives/edgar/data/1015650/000119312526210347/d90078d6k.htm

1.3 Material 6-Ks (event timeline)

Treasury-share disposals (comp-related, the 2026 price signal ₩50,276 → ₩56,860 → ₩100,000): 2026-01-28 https://www.sec.gov/Archives/edgar/data/1015650/000119312526025566/d53167d6k.htm ; 2026-02-04 https://www.sec.gov/Archives/edgar/data/1015650/000119312526036632/d159294d6k.htm ; 2026-04-27 (see table above).

Note: the PIPC ₩134.8B fine (2025-08-27) has no standalone 6-K; it is disclosed in the FY2025 20-F (provisionally paid; administrative appeal pending). Likewise the breach onset (public 2025-04-19) predates any 6-K; first EDGAR trace is 2025-05-02.

1.4 FPI adjustments (what does NOT exist)

• No 10-K / 10-Q / DEF 14A. SKM is a foreign private issuer; annual report = Form 20-F, interim reporting = Form 6-K. Governance/ownership disclosure comes from 20-F Items 6–7, not a proxy statement.
• Section 16 insider filings: N/A. FPIs are exempt; there is no Form 3/4/5 stream. Three anomalous Form 3/3/A filings exist (2024-12-23 https://www.sec.gov/Archives/edgar/data/1015650/000095017024139275/xslF345X02/ownership.xml ; 2026-02-19 https://www.sec.gov/Archives/edgar/data/1015650/000211225626000002/xslF345X02/primary_doc.xml ; 2026-02-20 https://www.sec.gov/Archives/edgar/data/1015650/000211225626000007/xslF345X02/primary_doc.xml) — all by Kim Taehee, “Investor Relations Officer,” disclosing a trivial 4,000-share holding, apparently voluntary/erroneous; no Form 4s exist. Disregarded as insider signal.
• Schedule 13D filed BY SKT (2024-12-20): $200M Penguin Solutions (PENG) convertible preferred via SPV Astra AI Infra LLC (6% PIK/cash, conversion $32.80784, 10.3% as-converted, 1 board seat; closed 2024-12-13). https://www.sec.gov/Archives/edgar/data/1015650/000095017024139216/
• Wellington Management 13G (2026-05-15): 11,598,806 sh = 5.4%, passive (event date 2026-03-31). https://www.sec.gov/Archives/edgar/data/1015650/000090221926000123/ ; earlier SC 13G/A 2022-02-14 https://www.sec.gov/Archives/edgar/data/1015650/000119312522039243/d270471dsc13ga.htm
• One Form SD (conflict minerals, 2022-05-20): https://www.sec.gov/Archives/edgar/data/1015650/000119312522155454/d300782dsd.htm

2. Quantitative cross-check feeds (third-party, reconciled to filings)

2.1 ROIC.ai aggregated data (statements, ratios, EV, multiples, price; pulled 2026-07-19)

Datasets pulled: company profile, income statement / balance sheet / cash flow (annual ×10, quarterly ×8, TTM), profitability, credit, liquidity and working-capital ratios, per-share data, enterprise value, valuation multiples, yield analysis, and latest price. Third-party aggregated data; reconciled to EDGAR XBRL (ifrs-full taxonomy, CIK 0001015650) and the FY2025 20-F.

What ties: revenue, net income, OCF, capex, dividends paid, cash, debt — exact match to EDGAR/20-F for FY2020–FY2025.

DATA-QUALITY WARNING (verified against the filings; do not propagate the aggregator’s figures):

Also: ROIC FY2020 mixes pre/post-restatement figures (restated continuing-ops revenue with an operating-income line matching neither basis). Same “EBITDA = op income” bug found in ROIC’s LG U+ feed (corrected with TTM D&A ₩2,717.7B before comping).

2.2 AZI (azitrading.com)

• Price CSV (https://azitrading.com/controls/download-data.php?t=SKM, pulled 2026-07-19): split/dividend-adjusted daily OHLCV back to the 1996-06-27 NYSE listing (7,561 trading days); latest $30.62 close 2026-07-17 (adjusted = unadjusted series). Basis for all price-action, EMA, beta/alpha, and return figures.
• azitrading.com valuation-percentile data (valuation_index, pulled 2026-07-19): DEGENERATE for SKM. latest shows P/E 15.77 / P/B 1.34 / P/S 0.93 with all three own-history percentiles identically 96.758 (composite 96.758, n_components 3) — but the history array is empty (length 0), so there is no empirical distribution to percentile against; the 96.758×3 is a feed fallback artifact, not a real percentile. The P/E level is also stale-denominator (AZI TTM EPS $1.94/ADS ≈ FY24 earnings vs true reported TTM ~$0.66/ADS, breach trough). Do not quote “96.8th percentile” as data.
• Rebuilt own-history method: multiples reconstructed by hand from 20-F/EDGAR fundamentals (EPS, book value, EBITDA, FCF) against the AZI daily price series at three anchor dates (Dec-2025 low $19.68, current $30.62, June-2026 high $46.00), giving the honest own-history read: P/B 1.29x now vs ~0.8–1.0x through 2022–2025; normalized P/E ~14.2x vs ~9–12x 2022–2025. The math is reproducible from the figures in Section 10.

2.3 FactorsToday factor API (pulled 2026-07-19; model date 2026-07-17)

Six endpoints pulled: stock loadings, leaderboard, stock info, stock-specific volatility, related stocks, and historical factor returns.

Key outputs: All-Factors model R² 0.194 (adj 0.168); dominant loading USDollar −0.54, Country: South Korea +0.32, Market +0.21, DividendYield +0.06; idiosyncratic vol 39.1% ann (~81% of variance not factor-explained); related-stocks are EM local-currency bond ETFs (LEMB/EBND/EMLC ~0.83–0.84) and international dividend/yield ETFs, Iberdrola, Banco Santander Chile — i.e., the model files SKM as a low-beta EM yield/KRW proxy, not with AI/tech names and not with telecom comps.

ADR caveats: (i) FactorsToday’s universe is US-listed securities; for an ADR of a KRW asset the USDollar/Korea loadings swamp the style read and loadings are estimated on US-hours returns that partly echo FX and overnight KRX moves — read them as “what the ADR trades as,” not the business’s factor DNA; (ii) leaderboard returns/Sharpe are annualized at every horizon (m3/m6 de-annualized and reconciled to the AZI CSV before use); (iii) R² 0.19 means most of the move is idiosyncratic anyway.

2.4 yfinance (pulled 2026-07-19)

• Short interest: 3.57M sh ≈ 1.35% of float, 1.1 days-to-cover (latest settlement; prior month 4.25M — declining). Third-party exchange data via yfinance, not a filing; flagged as such.
• FX: USDKRW 1,487.5 (2026-07-19).
• Comp pulls (KT 030200.KS, LG U+ 032640.KS, T, VZ, TMUS prices 2026-07-16/17) cross-checked against ROIC; LG U+ “EBITDA = operating income” bug corrected by adding TTM D&A ₩2,717.7B before computing EV/EBITDA.

3. News / press (material items validated at source)

Material items for which a URL was not captured (flagged — cite via the primary 6-Ks in Section 1.3 until re-sourced): CEO change announcement (Jung Jai-hun, SK Group reshuffle 2025-10-30 — corroborated by the 2026-03-26 AGM 6-K appointing him Representative Director); Samsung 6G/AI-RAN MoU (Nov-2025); Anthropic-stake revaluation rally (Asiae, 2026-01-22); KOSPI −9.99% circuit breaker (2026-06-23; sahi.com/Phemex); 15GW AIDC plan press coverage (PR Newswire 2026-07-04; Total Telecom 2026-07-06; RTTNews via stockanalysis.com ~2026-07-13 — primary: 6-K 2026-06-30 in Section 1.3); Korea Value-up program context (primary: Value-Up Plan 6-K 2024-10-24 in Section 1.3). KKR ~₩1T AIDC funding rumor is sourced primarily to the 2026-07-10 clarification 6-K (Section 1.3); follow-up disclosure due by ~2026-08-10.

4. Frameworks

• Frameworks applied:
  • Greenwald & Kahn, Competition Demystified — barriers-to-entry-first moat taxonomy; share-stability test (SKT passes: <2pp share movement over 5–8-yr windows, ~40 yrs as #1) and ROIC test (fails the 15–25% franchise threshold: ~5–6% ROIC 2022–24, 2.2% 2025); EPV vs growth-value framing used in the valuation split (telecom EPV vs embedded AIDC premium).
  • Marathon, Capital Returns — supply-side capital-cycle analysis; including the regulation-distorts-the-cycle application: Korean telecom returns never boom (state caps price) and never bust (state blocks entry) — the cycle is flatlined by design; and the capital-inflow warning applied to the simultaneous Korean AIDC build (all three telcos + Naver + Samsung SDS + state + hyperscalers, capacity landing 2027–29).

5. Reconciliation & caveats

Basis (applies to every figure in this report):

• KRW primary; consolidated K-IFRS/IFRS, FYE Dec 31. K-IFRS (domestic headline) vs IFRS-IASB (20-F) operating profit differs by reclassifications (FY25: K-IFRS ₩1,073.2B vs IFRS ₩1,048.5B; the ₩134.8B breach fine sits below the K-IFRS operating line in “Others” other expenses — SKT’s headline domestic op profit excludes it).
• ADR ratio: 1 ADS = 5/9 of one ordinary share — verified from the FY2025 20-F cover (Item 12.D). Ordinary price = ADR × 9/5 × FX; ADS gets ordinary × 5/9 of per-share economics.
• FX: USDKRW ~1,460 used for USD conversions (mid of the 1,443–1,488 range pulled across sources: SED 1,443 May-2026; yfinance 1,487.5 2026-07-19). KRW traded 1,322–1,588 over 52 weeks → ±~9% ADR move independent of fundamentals.
• Price discrepancy noted: ADR $30.62 (AZI close 2026-07-17, used as primary) vs $30.87 (ROIC snapshot, same date; +0.8%). KRX 017670 ~₩78,800–80,000 mid-Jul-2026 is consistent with parity at these inputs.
• Shares: 212,982,275 ordinary outstanding (214,790,053 issued − 1,807,778 treasury), 20-F 2025-12-31.

Data-quality caveats (consolidated; detail in Section 2):

1. ROIC FY25 “EBITDA” is operating income without D&A → its 7.25x net-debt/EBITDA and 19.9x EV/EBITDA are broken; true ~1.6x / ~4.4x.
2. ROIC FCF-yield “54.7%” is a sign error (OCF + capex); true FY25 FCF yield ~8–14% depending on cap basis.
3. ROIC per-share fields are per-ADS in KRW (×1.8 = per-ordinary).
4. ROIC operating income deviates from IFRS (FY24 +₩145B; FY25 −₩33B; FY23 +₩56B).
5. ROIC EV is keyed to a stale YE2025 price; all EVs/multiples in this report rebuilt at the live price.
6. AZI valuation_index is degenerate for SKM (empty history; 96.758×3 fallback; stale-denominator P/E) — rebuilt own-history used instead (Section 2.2).
7. AZI TrailingAnnualDividendYield prints a garbage 0.869 — do not quote.
8. Short interest (3.57M sh, 1.35% of float, 1.1 dtc) is third-party exchange data via yfinance, not a filing.
9. ~50 pre-2024 mirrored 6-K covers are boilerplate-only; substance sits in unmirrored exhibits — the 20-Fs cover those events retroactively.
10. FactorsToday treats SKM as a US common stock; ADR loadings are noisier than for a domestic filer (R² 0.19).

Accounting / one-timer flags:

• FY2021 net income includes a ₩1,147.6B discontinued-operations gain (SK Square spin-off, effective 2021-11-01; FY2020 includes ₩816.6B); SK Hynix 20.1% equity-method income (~₩1T+/yr) left the P&L from Nov-2021. Pre-2021 EPS restated for the 5:1 split.
• FY2025 effective tax rate 48.1% (vs 21–23% normal) — the ₩134.8B PIPC fine is non-deductible, plus subsidiary losses without tax benefit.
• FY2024: ₩94.7B impairments (incl ₩82.7B NATE/SK m&service, below K-IFRS op line); ₩207.9B discontinued-ops credit (Q4-24).
• Breach one-timers concentrated in H2-2025: fine ₩134.8B (Q3), ~₩500B Customer Appreciation Package (in revenue), USIM replacement costs, ~0.73M sub loss; Q3-2025 net loss ₩166.7B.
• Dividends: FY2025 DPS ₩1,660 (Q3-25 interim and FY25 year-end skipped) vs ₩3,540 FY2024; resumed ₩830 for Q1-2026.
• ADR-holder dividend economics: Korean withholding (~22% statutory incl. local surtax; 15% + local under the US treaty) plus depositary fees; USD payout carries full KRW translation risk.

No price targets. No recommendations. Facts, interpretations, and assumptions are labeled as such throughout.