Variant Perception

Where We Disagree With the Market

The variant view: the consensus mean target of ₩105,864 is not the operating-loss bear case it is sold as — the math only closes if ISC's KOSDAQ mark compresses roughly 25–30%, which no covering analyst has explicitly underwritten. SKC owns 45.03% of separately-listed ISC, worth ~₩2.42T against SKC's ₩6.05T market cap; once you mark that piece daily, the residual implied value of nexilis + picglobal + Absolics + SK leaveo less parent net debt is only ~₩1.6T at the consensus target — meaning either (a) consensus is implicitly pricing AI-multiple compression at ISC, or (b) the SOTP cannot be reconciled with the published targets. The market's stated debate is "rights-offering dilution + 1Q26 sustainability"; the actual decisive variable is ISC's daily standalone share price, which moves more SKC market cap per percentage point than any quarterly consolidated EBITDA print possibly can. We do not disagree that the stock is fully priced — we disagree about which observable signal resolves the debate, and we think the report's evidence base names that signal more honestly than any consensus note we found.

1. Variant Perception Scorecard

Reading the score. Consensus is unusually legible here — 9 sell-side analysts at mean UNDERPERFORM, ₩105,864 target, while spot is ₩159,900 and 30-day realised vol just printed 117%, the highest in a decade. That collision (uniformly bearish institutions, parabolic retail tape, ~22% rights dilution arriving 8 June) makes the positions easy to map. The variant strength is held below 70 because we agree with the bear's destination — we disagree with the bear's rationale. Resolution is short: the 8 June rights listing crystallises the dilution arithmetic; the 5 August 2Q26 EBITDA print rules on the operating inflection; ISC's daily KOSDAQ mark prints in the meantime. Four months is enough for all three to settle.

---

2. Consensus Map

The collision. The institutional bid is leaning on stale ₩90–105K targets; the tape is leaning on six-week +75% momentum; the company is days away from a ~22% share-count step-up. Each position implies a different model of the world. The variant view below sits inside that gap.

---

3. The Disagreement Ledger

Disagreement #1 — The decisive variable is ISC standalone, not 2Q26 EBITDA

A consensus analyst would say the August 5 consolidated EBITDA print resolves the SKC debate — does the +₩10B 1Q swing hold or revert. The report's own SOTP arithmetic disagrees: ISC's KOSDAQ mark (₩2.42T parent share, against SKC's ₩6.05T cap) moves more SKC market value per percentage-point shift than any plausible read of one quarterly EBITDA print. A 20% ISC compression strips ~₩484B from SKC's stake — roughly 8% of cap — without SKC management doing anything, while the consolidated EBITDA print resolves at most a ~₩30–80B operating swing. If we are right, the right thing for a PM to monitor first is KOSDAQ:095340 and the AI-capex revision wave at LEENO/Cohu, not the SKC investor-relations calendar. The cleanest disconfirming signal would be a 2Q26 EBITDA reversion to negative coinciding with ISC holding its mark within ±10% — i.e., the consolidated print does the work and the ISC mark does not.

Disagreement #2 — Consensus targets quietly require ISC to compress, but no analyst owns that view

A consensus analyst would say the ₩105,864 mean target reflects "operating losses, dilution, and below-cost-of-capital ROIC." The arithmetic does not survive that framing on its own. At ₩105,864 × ~37.9M pre-rights shares, market cap implied = ₩4.01T; minus the listed ISC stake at current KOSDAQ price of ₩2.42T leaves ₩1.59T for everything else. The SOTP residual — SK nexilis at replacement value, SK picglobal at trough peer multiples, Absolics + leaveo as options, less the parent net debt that sits outside ISC — does not credibly close to ₩1.59T unless ISC is also being marked down materially. The variant claim is that consensus targets are quietly an AI/HBM-multiple bet, not the operating bet they are framed as. If we are right, a PM short the stock on the operating thesis is also implicitly short the AI cycle through ISC — a position concentration that the published research has not made explicit. The cleanest disconfirming signal would be a covering analyst publishing an SKC note that explicitly flexes an ISC standalone scenario inside their target derivation; the cleanest confirming signal would be a sell-side downgrade arriving in lockstep with an ISC compression event, exposing the dependency.

Disagreement #3 — The "Q4 big-bath optical" critique does not touch the segment-level evidence

A consensus bear would say 1Q26 was flattered by the ~₩316.6B 4Q25 impairment in EV battery + chemicals — i.e., the QoQ comparison is mechanically wider than the underlying improvement. That is true at the consolidated line. It does not, however, explain a Malaysia copper-foil plant turning standalone EBITDA-positive (a plant-level operating fact, not a consolidated impairment artifact), NA copper-foil volume +403% YoY, ESS volume +390% YoY, ISC +236% YoY operating profit at 35% margin, or chemicals swinging to +₩9.6B operating profit first positive since 3Q22. Each of those is a segment- or plant-level print that the impairment wedge does not retroactively flatter. If we are right, 2Q26 EBITDA does not need to print +₩30B to validate inflection — it needs ISC to hold its margin and any one of (chemicals, Malaysia, NA pull-through) to hold its 1Q gain. The cleanest disconfirming signal would be Malaysia plant standalone EBITDA flipping back negative and chemicals reverting to a loss in the same quarter, with no offsetting ISC strength; that combination would validate the "1Q was an artifact" reading.

---

4. Evidence That Changes the Odds

---

5. How This Gets Resolved

The cadence. ISC's daily mark prints continuously and dominates the SOTP arithmetic — it is the only signal that requires no waiting. The 8 June listing is the next hard date and resolves the dilution arithmetic, after which the ₩105,864 consensus target either becomes mechanically reachable or visibly stale. The 5 August 2Q26 print closes the inflection debate. Sell-side revisions and Absolics customer announcements are slower and lumpier — they do not need to happen for the variant view to be tested.

---

6. What Would Make Us Wrong

The cleanest break of this view is a single combination: 2Q26 EBITDA reverts to a clean negative print and ISC standalone holds its current ₩5.37T mark with no compression. That combination would say (a) the inflection thesis was an accounting/PG-spread artifact after all, and (b) the SOTP held together precisely because the AI multiple is more durable than we credited. The bears would have been right about the operating story, and consensus targets would close mechanically through dilution alone — making our claim that "consensus is quietly an AI-multiple bet" look like reverse-engineered cleverness rather than a real edge.

A more ordinary way we could be wrong is on the SOTP arithmetic itself. Our variant view treats consolidated parent net debt as a deduction against the ex-ISC residual, and treats the ISC stake as if the public can dispose of it at the KOSDAQ mark. Neither is fully clean. If parent net debt allocation shifts (the rights-offering proceeds split is ₩590B Absolics / ₩410B debt repayment) the ex-ISC residual changes; if the market applies a holding-company discount to the ISC mark inside SKC (which it almost certainly does), the implied "everything-else" residual is larger than the gross arithmetic suggests, and the consensus target stops requiring ISC compression to close. We have priced both adjustments inside our ranges, but a careful PM should test the ranges independently before committing to the variant.

There is also a quieter way we could be wrong: the consensus may be perfectly aware that its targets require ISC compression and simply not say so in published notes. Korean sell-side is famously opaque on cross-affiliate scenario analysis, and the absence of an explicit ISC scenario does not prove the analyst is unaware of it. If that is the case, the variant view loses its information value — it is true but already priced. The test is whether sell-side revisions over the next 60 days reference ISC standalone explicitly. If they do, the variant has been incorporated; if they continue to reference only consolidated EBITDA and rights-offering dilution, the gap is real.

The one piece of upstream evidence we are most exposed to is the forensic file's flag on the ₩594B FY25 below-the-line wedge. We have argued that the wedge applies to the consolidated comparison and not to the segment-level inflection prints. That is true if the wedge is mostly equity-method losses (Absolics) plus finance costs plus FX. If the K-IFRS Korean filings reveal that a meaningful portion of the wedge is segment-level write-downs that should have been allocated to operating engines, the segment-level prints we cite get smaller and the bear's "accounting artifact" critique becomes more credible. The disaggregation lives in DART Korean filings and we have not seen it.

The first thing to watch is the daily KOSDAQ price of ISC (095340) — every other resolving signal in this tab is downstream of it, and it is the only one printing today.