A real moat, but the mispricing is already gone · −12% margin of safety
Market sits at the top of the MC distribution — P(intrinsic < market) = 95%Intrinsic value NT$164/share (after an 8% governance haircut; pre-haircut NT$179) vs market NT$186.50. The Monte Carlo p95 tail is NT$187 — today's price sits at the very top of 1,000 correlated draws, and 95% of them land below the market. This is the audit-trail reject: a genuine high-purity-solvent moat that the screen has already finished re-rating.
What it sells, where it sells
Operating segments
The electronic-grade mix (~76% and rising) carries a genuine moat — a 2-3 year build-and-qualify cycle and recycled-solvent purification where competitors barely exist. But ~40% of all sales route through a single customer (TSMC), and the industrial-grade tail swings with oil prices, so the consolidated number the market pays for is both concentrated and cyclical.
Country mix (revenue-weighted CRP input)
Revenue is ~82% Taiwan-based and follows the local foundry-fab supply chain, so the demand cycle is a TSMC node-ramp and semiconductor-capex story, not a diversified end-market. The Korea/China/SE-Asia tails (~18%) carry a small blended country-risk premium — but the dominant risk is single-customer concentration, not geography.
Background — five things to know before the case
Context the share price doesn't carry on its face. Skim this once and the bull / bear bullets below stop reading as inside baseball.
- The screen's "structurally accelerating earnings" thesis is only half-true. The PEG screen flagged 1773 as a sleepy chemicals name with accelerating earnings, but on the metric the reject note tracks, FY25 EPS actually fell ~6%, and the business is cyclical, not a secular compounder. At ~26x trailing P/E the easy mispricing the screen was hunting for is already gone.
- A real, hard-to-replicate moat — which is not the same as cheap. Of the world's top-10 foundries, all but the three Chinese ones are direct or indirect customers. Into TSMC, Shiny supplies 8 electronic-grade products — 4 as sole supplier and 3 as one-of-two — across 2nm-7nm nodes, with a 2-3 year build-and-qualification cycle and recycled-solvent purification capability where rivals "basically don't exist." The moat is genuine; it is also already in the price.
- Single-customer concentration is the dominant risk. TSMC alone was 38.2% of 2025 sales and crossed 40% in Q1'26. Earnings are tied to one foundry's node ramp and pricing — a structural fragility the 8% governance haircut only partly captures.
- Capacity expansions add depreciation before they add utilization. Two electronic-IPA lines (~NT$2.3bn capex) come online Sep 2026 and PGMEA expands from 5kt to 17kt — both self-funded from operating cash flow, with no financing pressure. But the new depreciation (from ~NT$670m/yr toward ~NT$800m/yr) pressures near-term gross margin on a see-saw until utilization climbs.
- The industrial-grade tail cuts both ways. ~24% of revenue is commodity, oil-price-linked solvent with immediate price pass-through. A Q2'26 raw-material spike flatters the print now, but the same line gives it back once feedstock costs normalize — volatility the market should not capitalize as growth.
Where we diverge from the sector
Each row asks: if we'd used the sector median for this one assumption instead of our override, how much would the share price change? Negative = our override is more conservative (worth less); positive = more aggressive (worth more). Skim down the rightmost column to see which bets are doing the work.
The story & the five claims
The 10-year story this DCF is built on, plus the five anchor claims that translate the story into engine inputs.
The 10-year story
Shiny Chemical is a real moat the disciplined DCF still says to walk away from. It makes high-purity electronic-grade solvents for the foundry chain — sole or dual supplier on multiple products into TSMC across 2-7nm nodes, with a 2-3 year qualification cycle that competitors cannot short-cut. That moat is genuine. What it is not is cheap: the screen flagged "structurally accelerating earnings," but FY25 EPS actually fell about 6%, the business is cyclical, ~40% of sales hang on a single customer, and at ~26x trailing P/E the easy mispricing the screen hunts for is already gone. The base case here is already generous — 9% near-term growth on the IPA and PGMEA capacity ramp, a 22% operating margin held just above the 2025 record, terminal growth pinned at Taiwan's 1.13% government-bond yield, all taken after an 8% governance haircut for the concentration and cyclicality. On those flattering inputs the company is worth NT$164 a share against a NT$186.50 price — a −12% margin of safety. At a normalized 9.4% stress WACC it is NT$89, −53%. The Monte Carlo puts the market at the 95th percentile of 1,000 draws: 95% of them land below today's price. This is the card the discipline is supposed to reject, and it does.
Two debates worth pressure-testing
- The inputs are deliberately generous, and it is still a SELL. We model 9% near-term growth and a 22% operating margin held above the 2025 record — the bull's own assumptions. On those flattering numbers the DCF is NT$164, a −12% margin of safety. The reject is not an artifact of conservatism; it survives the company's best case.
- FY25 EPS read as a fall, not an acceleration. The PEG screen anchored on a "structurally accelerating earnings" story; on the cyclical metric the reject note tracks, FY25 EPS fell ~6%, and ~24% of revenue is oil-price-linked commodity solvent the market should not capitalize as growth.
- β set to 1.05 by sector anti-suppression, not the 0.93 regression. Using the slightly suppressed regression β would only have flattered the WACC and the value. The sector-aware 1.05 lifts WACC honestly to 6.27% — and the verdict is still negative.
- Governance haircut 8%, heavier than the Taiwan peers. ~40% single-customer (TSMC) concentration and clear cyclicality earn a heavier 8% haircut than the 5% used on the more diversified Taiwan names; terminal ROIC faded to 10% to avoid an unfaded perpetual-moat assumption.
- Stress and Monte Carlo both confirm the reject. At a normalized 9.4% WACC the value is NT$89 (−53%), and 95% of 1,000 correlated draws land below the NT$186.50 price. The discipline rejected this card the way it rejected GOOGL and Bechtle — on the numbers, plainly.
- We could still be too generous. If 22% above-record margins prove unsustainable for a single-customer cyclical and revert toward the ~12% sector median, the DCF falls well below NT$164 and the overvaluation widens — the reject is conservative, if anything.
- Single-customer concentration is the real tail. TSMC at ~40% of sales means one foundry's node ramp, pricing, or in-sourcing decision moves the whole P&L. A concentration shock is not in the base case and is only partly priced by the 8% haircut.
- The capacity ramp see-saw could turn negative. New IPA/PGMEA depreciation (~NT$670m → ~NT$800m/yr) hits the P&L before utilization climbs. A slow ramp compresses gross margin precisely while the explicit-period growth is supposed to be strongest.
- The Q2 "war bonus" reverses. The industrial-grade book's raw-material spike flatters near-term prints; when feedstock costs normalize, the same pass-through gives the margin back. Capitalizing the spike as durable would be a mistake the cheap-looking trailing P/E invites.
- Even the bull case has limited headroom. To justify NT$186.50 you must underwrite both the full electronic-grade ramp and the low-rate TWD discount persisting — and the market already pays for both. There is no asymmetry left to capture.
Risks to thesis (tail, not bear case)
~40% of sales through one foundry. Any node-ramp slip, pricing renegotiation, or in-sourcing move hits the whole P&L. The dominant structural risk — only partly captured by the 8% governance haircut.
We hold 22% op margin above the 2025 record. If it reverts toward the ~12% Chemical (Specialty) median, intrinsic value falls well below NT$164 and the overvaluation widens.
New IPA/PGMEA depreciation (~NT$670m → ~NT$800m/yr) lands before utilization. A slow ramp compresses gross margin during the years the growth case needs most.
~24% of revenue is commodity, oil-linked solvent. The Q2'26 "war bonus" reverses once feedstock normalizes — volatility that should not be capitalized as growth.
A consumable product, but volumes still track foundry utilization and capex. A WFE downturn slows the very ramp the explicit period assumes. Cyclical, not structural.
The valuation leans on a 1.13% TWD risk-free; a move toward normalized rates compresses the perpetuity — and the explicit 9.4% stress WACC already lands NT$89, −53%.
10-year forecast
Revenue NT$11.5B → NT$21.8B over 10y (9% Y1-5 on the IPA + PGMEA capacity ramp, fading to 1.13% terminal). Operating margin lifts to 22% by Y5 and holds — just above the 2025 record 20.7%, a deliberately generous read that still leaves the DCF below market.
Monte Carlo distribution
The market sits at the 95th percentile of 1,000 correlated draws — only ~5% of outcomes clear today's NT$186.50, and the median draw (NT$162) is ~13% below price. The disagreement isn't how undervalued Shiny is; it's how confidently the discipline can say it's not a buy.
Mean NT$162.67 ± NT$13.77/sh, 1000 iterations (0 failed). P(intrinsic < market NT$186.50) = 94.8%.
Cost of capital build
| Risk-free rate | 1.13% |
| Mature-market ERP | 4.27% |
| Levered β | 1.05 |
| Weighted CRP | 0.91% |
| Cost of equity | 6.53% |
| Pre-tax cost of debt (synth Aaa/AAA) | 1.53% |
| D / V | ~5% |
| WACC | 6.27% |
Full year-by-year DCF
| Year | Revenue | Op mgn | EBIT | EBIT(1−t) | Reinvest | FCFF | PV |
|---|---|---|---|---|---|---|---|
| 1 | NT$12.54B | 20.95% | NT$2.63B | NT$2.14B | NT$1.29B | NT$844M | NT$794M |
| 2 | NT$13.66B | 21.21% | NT$2.90B | NT$2.36B | NT$1.41B | NT$949M | NT$840M |
| 3 | NT$14.89B | 21.47% | NT$3.20B | NT$2.60B | NT$1.54B | NT$1.07B | NT$889M |
| 4 | NT$16.23B | 21.74% | NT$3.53B | NT$2.87B | NT$1.68B | NT$1.20B | NT$939M |
| 5 | NT$17.70B | 22.00% | NT$3.89B | NT$3.17B | NT$1.83B | NT$1.34B | NT$991M |
| 6 | NT$19.01B | 22.00% | NT$4.18B | NT$3.39B | NT$1.64B | NT$1.75B | NT$1.22B |
| 7 | NT$20.12B | 22.00% | NT$4.43B | NT$3.58B | NT$1.39B | NT$2.19B | NT$1.44B |
| 8 | NT$20.98B | 22.00% | NT$4.62B | NT$3.72B | NT$1.08B | NT$2.64B | NT$1.64B |
| 9 | NT$21.55B | 22.00% | NT$4.74B | NT$3.81B | NT$709M | NT$3.10B | NT$1.82B |
| 10 | NT$21.79B | 22.00% | NT$4.79B | NT$3.84B | NT$304M | NT$3.53B | NT$1.96B |
Methodology & flags
Damodaran FCFF DCF, 10y explicit + perpetuity, in TWD. Risk-free
1.13% (local-currency government bond). Synthetic credit Aaa/AAA.
CRP 0.91% from revenue-weighted country mix × Damodaran 2026 CRPs.
Terminal ROIC faded to 10.00%; 8% governance haircut applied
post-DCF. Monte Carlo: 1000 iterations. Engine v1.0.0 · result:
/Users/valentin/Documents/notes_jiliac_labs/Finance/Damodaran/valuations/shiny/output/2026-05-31-result.json
- R&D cap: OFF · Lease cap: OFF · Failure: OFF · ESO: OFF
- Governance haircut: 8% applied post-DCF (NT$178.79 > NT$164.49)
- Sensitivity tornado: not run