Expected USD IRR

Expected USD IRR — 3-to-5-Year View

1. The Answer in One Paragraph

At today's price (₩27,450 close 2026-05-15 ≈ $18.39 per share at spot USD/KRW 0.00067), the probability-weighted 3-year USD IRR is roughly +2% and the 5-year USD IRR is roughly +7% — a thin reward that buries a wide range. The bull case (a Samsung HE-900IR extension plus a thin-film 양산 납품 order plus through-cycle revenue rebuilding toward long-term-thesis-claude.md FY30 path) prints +23% per year for five years in USD; the bear case (no second adjacency order, gross margin stuck below 50%, dilutive equity raise) prints -13% per year. The asymmetry is real but the central estimate sits below an equity hurdle rate — this is a watchlist IRR, not a buy IRR, until one of the four named catalysts in catalysts-claude.md market_focus lands and re-weights the distribution. FX adds ±2-3 points of USD IRR per ±10% KRW move and is the largest non-operating swing factor a USD investor faces.

2. Starting Position (USD)

Sources. Last close ₩27,450 on 2026-05-15 per data/prices/daily.json; diluted share count 9.233 M from numbers-claude.md share_count (FY25); market cap ₩253.5 B from deck.json cover stat; spot USD/KRW = 0.00067 from data/company.json fx_rates['2026-05-18']; FY25 net debt ₩4.5 B from numbers-claude.md net_debt. The USD market cap of ~$170 M and per-share price of ~$18.39 are both calculated at spot — they are the denominator in every IRR computed below. There is no analyst-coverage USD price target to compare against; the lone Korean-broker price target sits at ₩31,000 (variant-claude.md), or ~$20.77 at spot.

3. Scenarios at Exit (KRW → USD)

Three scenarios, each anchored to existing report tabs rather than re-derived. Bull and Bear take the 3-year exit from numbers-claude.md scenario_table and the 5-year exit from long-term-thesis-claude.md compounding_path (FY30 revenue and operating-margin path), with exit multiples held inside the cluster range from business-claude.md peer table. The Base 3-year price aligns with the lone covering analyst's ₩31,000 target from variant-claude.md.

A note on the bull's 3-year vs. 5-year exit price. The bull tab carries a ₩45,000 18-24 month price target (bull-claude.md); we lift the 3-year exit to ₩52,000 to reflect another year of FY28 compounding (revenue ₩85 bn × 11% op margin × 18× EV/EBIT plus a higher capitalised value on a second confirmed adjacency line) and the 5-year exit to ₩77,000 to reflect the full compounding_path FY30 print (₩105 bn revenue × 6.5× EV/Sales = ₩682 bn EV / 9.23 M diluted shares, plus modest net cash). This is the same engine as the bull-claude maths — extended one and three years forward and held to the same EV/Sales multiple. The Base 5-year price of ₩45,000 derives from the same FY30 ₩105 bn revenue base at a more conservative 4× EV/Sales (today's multiple, no cluster re-rating). The Bear price is the bear-claude.md floor (BVPS ₩6,874 × ~2.0× P/B); it does not improve over 3 vs. 5 years because the bear case assumes the cycle does not recover.

The bull case is flatter across horizons (23.7% over 3 years vs 22.9% over 5 years) because the 18-24 month target prices in a large fraction of the upside; the 5-year just extends compounding at the multiple. The base case widens on the 5-year (4.1% → 10.4%) because the compounding-path FY30 revenue base is materially higher than FY28 even at an unchanged exit multiple. The bear case narrows (-20.1% → -12.7%) because the same nominal floor amortised over more years is a smaller annualised loss — but the absolute drawdown is identical (-49% from entry to ₩14,000 exit). The scenario dispersion (≈45 percentage points of USD IRR between bull and bear over 3 years) is enormous for a single equity and is the structural feature that drives the watchlist verdict.

4. Probability Weighting

The weights below reflect the verdict-claude.md posture (Watchlist, bear-leaning until one of three catalysts lands), the variant-claude.md view of consensus, and the catalysts-claude.md calendar of dated trigger events inside the 12-24 month window. They are explicit, defensible, and the largest single judgment call in the IRR.

Doing the math: 3-year expected IRR = 0.25 × 23.7% + 0.45 × 4.1% + 0.30 × (−20.1%) = +1.75%. 5-year expected IRR = 0.25 × 22.9% + 0.45 × 10.4% + 0.30 × (−12.7%) = +6.60%. The 5-year number is higher because base-case compounding rebuilds more value as the underlying revenue base scales toward long-term-thesis-claude.md FY30 levels. Neither number clears a 10% USD equity hurdle on a probability-weighted basis — the expected IRR is informationally consistent with the verdict-claude "Watchlist" call.

5. Optional 4-Year Midpoint Sensitivity

Weighted 4-year IRR = 0.25 × 24.0% + 0.45 × 8.5% + 0.30 × (−15.5%) = +5.2%. The 3-yr / 4-yr / 5-yr arc (1.75% → 5.2% → 6.6%) shows the expected return rebuilds gradually as base-case operating leverage compounds, but the bear case caps the path until the binary event resolves.

6. FX Sensitivity

The IRRs above all use spot USD/KRW = 0.00067 (≈ ₩1,492/USD) held constant. KRW has weakened progressively over the past two years (data/company.json fx_rates: 0.00084 at 2021-12-31 → 0.00068 at 2024-12-31 → 0.00067 today), and the forward path is unhedged in USD. The table below holds operating scenarios constant and varies exit FX.

Read. A ±10% KRW move shifts the weighted 5-year USD IRR by roughly ±2.3 percentage points — meaningful, but smaller than the operating-scenario dispersion (±15-18 points around the mean). Said differently: about three-quarters of the USD IRR variance is operating, one-quarter is FX. A USD investor cannot ignore FX (a 10% sustained KRW devaluation would knock the weighted 5-year IRR down to ~4%), but cannot hide behind it either — the dominant return driver is whether the adjacency option converts.

7. What Has To Be True

The IRR distribution above resolves to a single number for each scenario only if a small set of named milestones land or fail. The mapping below ties each scenario to the specific dated events in catalysts-claude.md and long-term-thesis-claude.md that have to print.

8. Risks to the USD IRR

KRW devaluation is the largest non-operating risk. USD/KRW has weakened ~20% from 2021-12-31 (0.00084) to 2026-05-18 (0.00067) per data/company.json fx_rates. A USD investor who buys the equity at ₩1,492/USD and exits at ₩1,667/USD loses ~10% of USD return independent of any operating outcome — and the rate is unhedged. The FX-sensitivity table above bracket-tests ±10% and shows ±2.3 points of USD IRR; a tail scenario of a sustained 20%+ KRW weakness (driven by Korean trade balance deterioration or a Bank of Korea easing cycle) would meaningfully erode the central case. The IRR is not USD-defensible without FX.

Equity-dilution risk per bear-claude.md is the operating tail. FY25 net debt is ₩4.5 bn, quick ratio 1.04×, and short-term borrowings are 88% of total debt — a second consecutive loss year forces either a primary equity raise (lowers per-share value at any given enterprise value — IRR mechanics are unchanged at the EV level but the per-share path is steeper) or an R&D cut (slows the cadence that is the moat). Neither outcome appears in the scenarios above; both would shift the realised IRR closer to the bear case at a given enterprise value. The bear scenario already implicitly bakes a dilution outcome into the ₩14,000 BVPS-floor exit.

Exit-multiple risk is asymmetric across scenarios. The bear case uses a 2.0× P/B floor (FY22 trough) because EV/Sales becomes meaningless on a loss-making revenue base — there is no operating cushion to multiply. The base and bull cases use 4-6.5× EV/Sales because by FY28-FY30 revenue should be cycle-recovering and the cluster comparison is the relevant frame. This is methodologically correct but unstable — if the bull case is right on revenue but the market refuses to extend cluster multiples (because Park Systems' 9.4× P/B is the existence proof, but not the rule), bull-case IRR compresses by 5-8 points. The IRR is robust to operating outcomes; it is fragile to multiple-regime changes.

9. Verdict

The probability-weighted USD IRR of ~2% (3-year) to ~7% (5-year) does not justify the stock at today's price for a USD investor with a typical equity hurdle rate. The asymmetry is real — the bull case clears 20%+ USD if the adjacency converts — but the central case sits inside the noise band of cost of capital, and the bear case loses double-digits annually. This is a watchlist IRR, not a buy IRR; the trade is to size into the position only on confirmation of the next observable catalyst (Samsung HE-900IR extension, SK Hynix qualification, or thin-film 양산 납품 order), which re-weights the distribution and pulls expected USD IRR above 10%.