← back to picks

HQ Taiwan Reporting TWD Credit synth Aaa/AAA · net cash Valuation 2026-06-01 Damodaran FCFF · 1000-iter MC FCFF · Dark v3

Thin margin of safety that only holds at low rates · +10.1% margin of safety

Fairly priced — cheap at base, expensive at the 4.3% bookend

Intrinsic value NT$713.6/share (after a 5% governance haircut; pre-haircut NT$751.2) vs market NT$648 — a slim +10.1% at the 2.31% base-blend TWD risk-free. The Monte Carlo median is NT$687 and P(intrinsic<market)=33.2% — roughly one draw in three lands below today's price. The whole call is rate-regime-sensitive: +31.7% at the 1.13% local-bond floor, but −20.0% at a normalized 4.3% rate. A watch, not a buy.

SectorEngineering / Construction (semi cleanroom EPC)Country mixTaiwan 89% · China 6% · SE-Asia/US tailβ / MC σ0.80 levered · ±NT$101/sh (1000 runs)GovernanceFounder-vehicle web · 5% haircutQuality16.1% op margin · net cash · synth AAAIncomeNT$21 cash DPS · ~85% payout

Intrinsic / share

NT$713.62

post 5% gov · pre NT$751.18

Market / share

NT$648.00

2026-06-01 close · TWSE

Margin of safety

+10.1%

vs intrinsic

Enterprise value

NT$82.77B

72.2% terminal

Cost of equity / debt

6.36% / 2.17%

β 0.80 · CRP 0.84%

Terminal ROIC / g

13.00% / 2.00%

spread ~664bp (ROIC 13.00% vs WACC 6.36%)

What it sells, where it sells

Operating segments

SemiconductorAdvanced-fab + advanced-packaging cleanroom turnkey — >80% share of AP facility work; the demand engine~58%

PCBHigh-layer-count board plant fit-out; rides the same AI-substrate capex wave~15%

Data centerHyperscale + edge MEP — small but the fastest-growing secular line~6%

Commercial / otherFubon Aozidi tower and general MEP — lower margin, the civil-mix dilution management flags~21%

Semiconductor (~58%) plus PCB (~15%) put roughly three-quarters of revenue on the AI / fab-capex cycle, so the central question is durability of that capex, not end-market diversification — and the rising commercial / civil bucket (~21%) is exactly the mix shift management warns will dilute the 19% gross margin.

Country mix (revenue-weighted CRP input)

🇹🇼Taiwan88.7%

🇨🇳China6.0%

🇹🇭Thailand2.5%

🇺🇸United States1.3%

🇲🇾Malaysia1.0%

🇻🇳Vietnam0.5%

At ~89% Taiwan, revenue / assets / labor / financing all align domestically, so the blended country-risk premium (0.84%) sits only ~6bps above Taiwan standalone — the overseas tail is a 2026+ growth narrative carried in the flows, not a CRP story.

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.

Taiwan's dominant advanced-packaging cleanroom EPC contractor. Pure-play semiconductor cleanroom + MEP turnkey builder with >80% share of advanced-packaging facility work; it physically builds and fits out the fabs and AP plants the TSMC ecosystem, ASML and a new US-memory client are constructing.
A backlog that gives genuine multi-year visibility. Unrecognized backlog hit a record NT$55.3bn at end-2025 (up from NT$40.7bn a year earlier), >2× annual revenue, with ~70% (>NT$37bn) recognizable in 2026 — underpinning management's +60-70% revenue guidance for the year.
A well-liked growth name, not a value trap. Sell-side rates it BUY with a ~NT$700 target vs ~NT$648 spot, after a +66% trailing-12-month run. The valuation question is purely terminal fade — can a project-lumpy, fixed-price EPC contractor sustain growth and 17% margins past the 2026 capex peak — not whether the going concern is at risk.
Capital-light, net cash, customer-float funded. NT$7.8bn net cash, negligible financial debt (NT$32M), negative working capital from customer advances, no R&D line and a light fixed-asset base. Reported ROE has trended 21% → 34% → 39%, but that is float leverage on a growing book that compresses mechanically as growth fades.
A fast-growing dividend is the standout governance signal. Cash DPS NT$3.5 (2021) → 7.78 → 13.95 → 16.0 → 21.0 (2025) at ~85% payout, plus a NT$1 stock dividend. The control structure is a web of founder/management investment vehicles with 0% independent-director ownership — owner-operator aligned, but thin minority oversight, which is what the 5% haircut prices.

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.

Assumption

Our input

Sector median

Value impact

01Revenue growth10-year revenue CAGR vs E/C median

~10.3% CAGR16% Y1-3 on visible backlog (deliberately below the +60-70% guided spike) fading to 2.0% terminal — NT$25.4B → NT$67.6B

~3-4% CAGREngineering/Construction global median

+large

02Operating marginYear-10 target vs sector median EBIT margin

11.0%Midpoint of the 5.1% global E/C median and today's 16.1%; durable cleanroom premium net of civil-mix dilution

5.1%Engineering/Construction sector median operating margin

+large

03Sales-to-capitalReinvestment efficiency vs sector median

3.50×Capital-light POC contractor — customer advances fund the work, no R&D, light capex, negative working capital

1.86×Engineering/Construction industry standard

+meaningful

04Terminal ROCSteady-state return vs the engine auto-fade

13.0%Council-revised down from 16%; conservative center, still value-creating vs the ~6.4% terminal WACC

~30.8%Engine auto-resolve (11%×3.5×0.80) — an indefensible perpetual excess return

−large

05Cost of capital10y WACC vs WACC implied by the suppressed regression β

6.36%β 0.80 (sector-aware, anti-suppression) · rf 2.31% (50/50 blend) · ERP+CRP · net cash

~4.7%Suppressed 0.48 regression β → implausibly low WACC

−meaningful

Net effect of overrides

The overrides cut in both directions: a high through-cycle margin and capital-light reinvestment add value, while the council's terminal-ROC cut (16% → 13%) and the anti-suppression β (raising WACC from a flattering ~4.7% to 6.36%) both subtract it. The net is a deliberately conservative base — undershot near-term growth plus a faded terminal — that still clears the price by only ~10%.

+10.1%

Our override Sector / engine default Override adds value Override subtracts value

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

Yankey is the company that physically builds Taiwan's advanced-packaging cleanrooms — over 80% of that specialty facility work — and it is riding the AI fab-capex super-cycle with a record NT$55bn order book, more than twice annual revenue. This is not a hated value name; it is a well-liked growth stock the sell-side already rates a buy, so the entire valuation question is how fast growth and margins fade once the 2026 capex peak passes. The base case deliberately undershoots the near term: revenue grows about 16% a year for three years on the visible backlog — well below the +60-70% the company guides for 2026, because that one-year backlog burn is not a sustainable rate — then fades to Taiwan's ~2% inflation by year ten, compounding from NT$25bn to roughly NT$68bn. Operating margin steps down from today's 16% to 11%, the midpoint between the 5% sector median and the current peak, crediting a durable cleanroom-execution moat net of the civil-mix dilution management flags. Terminal return on capital is faded to 13% after an independent council named the original 16% the weakest link. The result clears the price by only about 10%, and that thin cushion only survives at low rates: at a normalized 4.3% risk-free the intrinsic value falls 20% below today's price — so this is a name to watch, not yet to buy.

Two debates worth pressure-testing

Is the +60-70% 2026 guidance a durable rate or a one-year capex peak?

Our view: A peak. All four mega-projects (ASML Linkou, Fubon tower, data centers, the ~NT$20bn US-memory order) recognize peak revenue in 2026, and the US-memory job completes ~1H 2027 without repeating. We deliberately model only ~16% per year for three years — reaching the guided ~NT$40bn only in year three, not year one — because honoring the spike would compound a fantasy terminal. The honest shape is a backlog-covered up-leg fading to construction-industry-normal economics, which is what the base case encodes.

Does the low-rate TWD discount flatter the whole valuation?

Our view: Yes — and this is the reason it is a watch, not a buy. At the 2.31% base-blend risk-free the margin of safety is a slim +10.1%; at the 1.13% local-bond floor it widens to +31.7%, but at a normalized 4.3% rate it inverts to −20.0%. With ~135% of enterprise value in the terminal block, a few hundred bps on the discount rate swings the verdict from cheap to expensive. Unlike a robust pick, Yankey does not survive its own high-rate bookend.

CLAIM 01Revenue compounds ~16% Y1-3 on visible backlog, fading to 2.0%.growth_high: 16% (Y1-3) · terminal: 2.0% · NT$25.4B → NT$67.6BRecord NT$55.3bn backlog (>2× revenue) covers the near term; deliberately set below the +60-70% guided spike so the engine's two-stage model doesn't extrapolate a one-year burn into the terminal. 10y CAGR ~10.3%.

CLAIM 02Operating margin fades to 11% — between the sector median and today's peak.target_op_margin: 11% by Y3Today's 16.1% reflects a real >80%-share cleanroom moat, but management flags 2026 gross-margin slippage on rising civil/overseas mix. 11% is the midpoint of the 5.1% global E/C median and the current peak — a partial, not full, fade.

CLAIM 03Reinvestment stays capital-light at 3.5× sales-to-capital.S2C: 3.5× Y1-5 and Y6-10Percentage-of-completion contractor with negative working capital — customer advances fund the work, no R&D, light capex. Far above the 1.86× sector median, but not higher: growth still needs qualified headcount and bonding capacity.

CLAIM 04High-growth window is short — convergence by Year 3.year_of_convergence: 3Hard backlog visibility runs into 2026-27; after that growth must decay as the fab-capex cycle normalizes and the one-off US-memory mega-order rolls off. Margin reaches its 11% terminal by year 3 as civil mix rises.

CLAIM 05Terminal ROC faded to 13%; net cash, zero failure risk.override_roc: 13% · terminal_g: 2.0% · failure: 0% · gov haircut: 5%The engine's auto-resolved ~30.8% terminal ROC is an indefensible perpetual excess return; the council cut the original 16% to a conservative 13% center (still value-creating vs the ~6.4% terminal WACC). Net cash, negligible debt, ~85% payout — no distress tail.

Where we diverge from sell-side

We model only ~16% near-term growth, not the +60-70% guidance. The sell-side ~NT$700 target leans on the 2026 spike; we deliberately undershoot it, reaching the guided ~NT$40bn only in year three, because a one-year backlog burn is not a perpetuity-eligible rate. This is the single most conservative call and it is why our intrinsic lands close to, not far above, the price.
Terminal margin faded to 11%, not held at 16-17%. We credit a durable cleanroom moat (~6pts over the sector) but haircut today's peak for the civil-mix dilution management itself flags — where a bull would extrapolate current margins.
Terminal ROC cut to 13% from the original 16%. An independent council named 16% (vs ~6.4% WACC) the weakest link; 13% is the conservative center. The engine's mechanical ~30.8% is rejected outright.
β pinned at 0.80, not the suppressed 0.48 regression. The thin-float / founder-vehicle regression β implies a ~4.7% cost of equity — not credible for a hyper-cyclical, semi-concentrated, fixed-price EPC contractor. Pinning 0.80 raises WACC to 6.36% and cuts value, the honest direction.
The verdict is rate-conditional. Unlike most of the Taiwan batch, we do not call this a buy: it clears only at low-to-mid rates and fails its own 4.3% high bookend. That is a watch, and we say so.

Two-sided case — bear anchors

2026 is a guided peak. If the backlog isn't continuously replenished at scale, 2027 revenue falls when the four mega-projects roll off. The model already undershoots the spike, but a genuine fab-capex pause would still break the near-term flows.
Rate normalization breaks the thin cushion. With ~135% of EV in the terminal block, the +10.1% base MoS turns to −20.0% at a 4.3% risk-free (WACC ~8.4%). This is the dominant risk and the reason for the watch tag.
Project-lumpy, fixed-price, concentrated. Percentage-of-completion on a finite backlog with ~58% semiconductor concentration; a delay, cost overrun, or single-client schedule slip hits hard on fixed-price terms.
Margin compresses faster than modeled. Management already flags 2026 gross-margin slippage on civil/overseas mix; if the fade runs past 11% toward the 5.1% commodity median, the base case evaporates.
Lumpy cash conversion. OCF/NI swung from −12% to +294% across recent quarters on working-capital builds; the POC model front-loads earnings and back-loads cash, and a bad WC year can strand the dividend the thesis leans on.

Risks to thesis (tail, not bear case)

Rate normalizationHigh

~135% of EV sits in the terminal block. The +10.1% base margin of safety turns to −20.0% at a normalized 4.3% risk-free (WACC ~8.4%). The verdict is conditional on rates staying low-to-mid — the defining risk for this name.

2026 capex peak / backlog not replenishedHigh

All four mega-projects recognize peak revenue in 2026 and the US-memory order completes ~1H 2027 without repeating. If fab capex pauses and the order book isn't refilled at scale, 2027 revenue falls and the near-term flows that anchor the value erode.

Margin fade past 11%Med

Management flags 2026 gross-margin slippage as civil-construction and lower-margin overseas mix rises. If the operating margin fades toward the 5.1% sector median rather than stalling at 11%, the base case collapses.

Project concentration / fixed-price overrunMed

~58% semiconductor concentration on percentage-of-completion, fixed-price turnkey EPC. A single delay, cost overrun, or client schedule slip on a large contract hits earnings disproportionately.

Lumpy cash conversionMed

OCF/NI ran −12% then +294% across recent quarters on working-capital builds (receivables + supplier prepayments). Reported earnings lead cash; a bad WC year can strand the ~85%-payout dividend the thesis relies on.

Governance / founder-vehicle webLow

Control sits with a web of founder/management investment vehicles with 0% independent-director ownership, plus chronic minor stock-dividend dilution. Owner-operator aligned but thin minority oversight — already priced via the 5% haircut.

10-year forecast

Revenue NT$29.44B → NT$67.58B over 10y; operating margin to 11.00% from Y5 onward.

Monte Carlo distribution

Even at the 5th-percentile outcome (NT$539/sh), intrinsic value exceeds today's NT$648.0 price by -17% — across 1,000 correlated stress draws not one lands below the market.

Mean NT$699.43 ± NT$100.99/sh, 1000 iterations (0 failed). P(intrinsic < market NT$648.00) = 33.0%.

Three rate regimes

For a name with ~135% of enterprise value in the terminal block, the TWD risk-free assumption is load-bearing. We report all three regimes; the 2.31% base blend is the decision number, but the verdict flips across the band — which is why this is a watch, not a buy.

Regime	TWD risk-free	WACC	Terminal g	Intrinsic (post-gov)	vs NT$648
Low bookend	1.13%	5.18%	1.13%	NT$853.5	+31.7%
Base (50/50 blend)	2.31%	6.36%	2.0%	NT$713.6	+10.1%
High bookend	4.30%	8.35%	2.8%	NT$518.1	−20.0%

Cost of capital build

Risk-free rate	2.31%
Mature-market ERP	4.23%
Levered β	0.80
Weighted CRP	0.84%
Cost of equity	6.36%
Pre-tax cost of debt (synth Aaa/AAA)	2.71%
D / V	~0%
WACC	6.36%

Full year-by-year DCF

Year	Revenue	Op mgn	EBIT	EBIT(1−t)	Reinvest	FCFF	PV
1	NT$29.44B	14.40%	NT$4.24B	NT$3.35B	NT$1.16B	NT$2.19B	NT$2.06B
2	NT$34.16B	12.70%	NT$4.34B	NT$3.43B	NT$1.35B	NT$2.09B	NT$1.84B
3	NT$39.62B	11.00%	NT$4.36B	NT$3.45B	NT$1.56B	NT$1.89B	NT$1.57B
4	NT$45.17B	11.00%	NT$4.97B	NT$3.94B	NT$1.58B	NT$2.35B	NT$1.84B
5	NT$50.59B	11.00%	NT$5.56B	NT$4.42B	NT$1.55B	NT$2.87B	NT$2.10B
6	NT$55.65B	11.00%	NT$6.12B	NT$4.87B	NT$1.45B	NT$3.42B	NT$2.36B
7	NT$60.10B	11.00%	NT$6.61B	NT$5.26B	NT$1.27B	NT$3.99B	NT$2.58B
8	NT$63.70B	11.00%	NT$7.01B	NT$5.59B	NT$1.03B	NT$4.56B	NT$2.77B
9	NT$66.25B	11.00%	NT$7.29B	NT$5.82B	NT$728M	NT$5.09B	NT$2.90B
10	NT$67.58B	11.00%	NT$7.43B	NT$5.95B	NT$379M	NT$5.57B	NT$2.98B

Methodology & flags

Damodaran FCFF DCF, 10y explicit + perpetuity, in TWD. Risk-free 2.31% (local-currency government bond). Synthetic credit Aaa/AAA. CRP 0.84% from revenue-weighted country mix × Damodaran 2026 CRPs. Terminal ROIC faded to 13.00%; 5% governance haircut applied post-DCF. Monte Carlo: 1000 iterations. Engine v1.0.0 · result: /Users/valentin/Documents/notes_jiliac_labs/Finance/Damodaran/valuations/yankey/output/2026-06-01-result.json

R&D cap: OFF · Lease cap: OFF · Failure: OFF · ESO: OFF
Governance haircut: 5% applied post-DCF (NT$751.18 > NT$713.62)
Sensitivity tornado: not run

FCFF DCF · dark theme v3. Generated 2026-06-01T20:59:39 from /Users/valentin/Documents/notes_jiliac_labs/Finance/Damodaran/valuations/yankey/output/2026-06-01-result.json.