Screened: 2026-05-28 | Source: yfinance, SECON Alpha, StockAnalysis
Type: ABBREVIATED (YTD +130%, price within 9.1% of 52-week high)
| Metric | Value |
|---|---|
| Current price | TWD 927 |
| 52-week range | TWD 278.5 – 1,020 |
| YTD return | +130% |
| Distance from 52w high | -9.1% |
| 50-day avg | TWD 584.9 |
| 200-day avg | TWD 471.3 |
| Market cap | TWD 443B (~USD 13.6B) |
Abbreviated trigger met: YTD > 80% AND within 10% of 52-week high. The cycle has already inflected hard — this is not a trough entry.
The stock was at TWD 278 twelve months ago (deep trough). It has since tripled on a combination of: (1) sector sentiment recovery around AI semiconductor demand, (2) USD $406M US Commerce Dept grant for GlobalWafers' Texas plant, (3) management's Q1 2026 commentary that "Q1 may be a relatively low point in the current cycle." The stock moved before fundamentals confirmed the recovery.
| Metric | Value | Graham Threshold | Signal |
|---|---|---|---|
| Trailing P/E | 60.8x | ≤ 15x | FAIL |
| Forward P/E | 33.5x | ≤ 15x | FAIL |
| P/B | 4.75x | ≤ 1.5x | FAIL |
| P/E × P/B | 289 | ≤ 22.5 | FAIL |
| EV/EBITDA | 30x | — | Expensive |
Graham IV break-even growth rates:
At g=10%, Graham IV on forward EPS = TWD 788, still 15% below current price of TWD 927. The stock is not defensible on Graham's framework at any plausible growth assumption.
| Year | Dividend (TWD) | Change |
|---|---|---|
| 2022 | 16.0 | — |
| 2023 | 16.0 | flat |
| 2024 | 19.0 | +19% (peak earnings year) |
| 2025 | 11.0 | -42% cut |
| 2026 (declared) | 7.7 | tracking lower again |
The dividend is not stable. It was cut 42% in 2025 (from TWD 19 to TWD 11) as earnings fell from peak. 2026 declared dividend of TWD 7.7 represents another ~30% reduction. The trailing yield at TWD 927 is only 0.83% — irrelevant as an income source. Payout ratio is 85% of a declining net income base, with FCF deeply negative, meaning dividends are being partially funded by debt.
This definitively fails the Graham dividend stability criterion.
FCF has been negative for 3 consecutive years due to a massive capex supercycle:
| Year | CapEx (TWD B) | OCF (TWD B) | FCF (TWD B) |
|---|---|---|---|
| 2022 | -12.4 | +37.6 | +25.2 |
| 2023 | -36.8 | +18.6 | -18.2 |
| 2024 | -48.3 | +15.0 | -33.3 |
| 2025 | -33.5 | +12.7 | -20.8 |
The company is building a USD 5B fab in Sherman, Texas (partially financed by the CHIPS Act grant). This is strategic but consumes all operating cash flow and more. Net debt rose to TWD 54.8B vs equity of TWD 93.3B. D/E (total) = 0.72x; Net Debt/Equity = 0.59x. Current ratio is 1.24 (below Graham's 2.0 minimum), quick ratio 0.55 (very thin).
Book value per share = TWD 195 vs price TWD 927 → P/B 4.75x. Book value is not defensible as a floor.
| Filter | Criterion | Result | Status |
|---|---|---|---|
| 1. Size | Mkt cap > $2B | USD 13.6B | PASS |
| 2a. Current ratio | ≥ 2.0 | 1.24 | FAIL |
| 2b. LT debt | ≤ 2× working capital | TWD 43B vs TWD 65B limit | PASS (borderline) |
| 3. Earnings stability | Positive EPS every year 10yr | Listed 2014; profitable throughout incl. 2025 | PASS* |
| 4. Dividend record | Uninterrupted ≥ 20 yr | Listed 2014 only; 2025 cut -42% | FAIL |
| 5. Earnings growth | +33% over 10yr | EPS declining: TWD 35 (2022) → TWD 15 (TTM) | FAIL |
| 6. P/E | ≤ 15x | 60.8x trailing / 33.5x forward | FAIL |
| 7. P/B | ≤ 1.5x | 4.75x | FAIL |
Score: 2/7 passes (size + borderline LT debt). Everything valuation-related fails by a wide margin.
*EPS stability passes on no-loss criterion but conceals a 57% peak-to-trough EPS decline.
This situation resembles the Maersk pattern warned in shr-017: high trailing screen scores (for Maersk) came from a cycle peak; the stock now re-rates after earning guidance cuts confirmed a downcycle. For GlobalWafers, the opposite concern applies: the stock has already re-rated UP from a trough, pricing in recovery before it shows in financials.
The risk here is the inverse of shr-017: the market has already priced in the cycle recovery, buying at cycle trough; now fundamentals need to catch up to the 33.5x forward P/E. If the recovery is slower or shallower, there is significant downside.
| Category | Signal | Rating |
|---|---|---|
| Recent price action | +130% YTD, at 52w high | RED — no margin of safety |
| Earnings trend | 4 consecutive quarters of declining EPS | RED |
| Q1 2026 vs estimates | -25.4% miss | RED |
| Dividend | Cut 42% in 2025, tracking lower 2026 | RED |
| FCF | Negative 3 consecutive years | RED |
| Analyst consensus | Hold; mean PT TWD 659 vs price TWD 927 | RED — 29% downside to consensus |
| Analyst PT range | Low TWD 470, High TWD 901 | Even the HIGHEST analyst PT is below current price |
| Book value | P/B 4.75x | RED |
| Capex burden | USD ~$1.5B/yr ongoing US plant | RED (balance sheet pressure) |
| Insider activity | No Finnhub data available (TWO listed) | NEUTRAL |
| Portfolio overlap | Silicon wafer supply chain — less AI-correlated than TSMC | AMBER |
All-RED pre-buy check. The stock currently trades ABOVE every analyst price target. The consensus mean (TWD 659) implies 29% downside from current levels. This is not a value entry — it is post-recovery speculation.
You hold AGN.AS / RI.PA / AIG / AKE.PA. The original thesis was: "less AI-correlated than HBM-substrate names." That was true at TWD 278 (the trough). At TWD 927, GlobalWafers has already rallied harder than most AI-correlated semis YTD. The thesis's core value proposition (buying at a discount to intrinsic value during a trough) has been captured by the market — you'd be entering at the other end of the cycle.
VERDICT LINE: RED. Do not buy at current levels.
The abbreviated trigger was met because the cycle has already inflected. GlobalWafers at TWD 927 fails 5 of 7 Graham filters, trades at 60.8x trailing / 33.5x forward P/E, P/B 4.75x, FCF has been negative for 3 years, the dividend was cut 42% and is still declining, and every analyst's price target sits below the current price (consensus TWD 659, high PT TWD 901). There is no margin of safety at any plausible Graham growth assumption.
Watchlist entry price: TWD 370–420 (prior to the AI-sentiment rally, around P/B ~1.8-2.0x and forward P/E ~15-18x at trough earnings). That level would represent a 55–60% drawdown from current price — only warranted if the capex supercycle creates a genuine oversupply / earnings bust scenario.
Re-screen trigger: Price below TWD 420 with confirmed dividend stability (two consecutive payments without cuts) and FCF turning positive (capex normalizing post-Texas fab completion, expected 2026–2027).