Screened: 2026-05-28 | ADR price: $66.05 | Underlying: 8150.TW at TWD 103 | FX: 31.36 USD/TWD
CRITICAL CONTEXT: The thesis hook described a trough valuation at 7%+ yield and sub-book price. That trough no longer exists. IMOS is trading at its 52-week HIGH today (+16.4% intraday), up +116% YTD. This screen is evaluating the stock at its current price, not the hypothesized entry.
| Metric | Value |
|---|---|
| Price (ADR, USD) | $66.05 (+16.4% today) |
| 52-week range | $15.09 – $66.43 |
| YTD return | +116.6% |
| 5-year return | +171.5% |
| Market cap | ~$2.30B USD |
| ADR ratio | 1 ADR = ~20.3 TWD shares (8150.TW) |
| Short % of float | 0.14% (negligible) |
The stock hit its 52-week high intraday on the screen date. The +340% move from the Q2 2025 trough (~$15) has already happened.
All EPS figures are USD per ADR, derived from TWD financials via FX conversion and ADR ratio. The yfinance-reported trailing EPS ($0.80) is used as a sanity check; Finnhub quarterly actuals provide the more reliable quarterly decomposition.
| Multiple | Value | Note |
|---|---|---|
| Trailing P/E (LTM 4Q) | 56x | EPS $1.18 (Q2-25 to Q1-26) |
| Forward P/E (Q1 annualized) | 23x | EPS $2.88 |
| P/B | 2.93x | BVPS $22.58 |
| P/S (2025 revenue) | 3.0x | |
| EV/EBITDA (2025) | 12.5x | |
| Dividend yield | 1.2% | $0.82 last div; was 7%+ at $15 trough |
| Payout ratio (2025) | ~163% | Paying more than earned |
| Item | TWD M | USD M |
|---|---|---|
| Cash & equivalents | 12,390 | 395 |
| Total debt | 15,820 | 504 |
| Net debt | 2,570 | 82 |
| Current assets | 24,340 | 776 |
| Current liabilities | 9,990 | 319 |
| Working capital | 14,350 | 458 |
| Shareholders equity | 24,610 | 785 |
| Current ratio | 2.44x | — |
| Interest coverage | ~3–4x | EBIT / interest expense |
Net debt is modest ($82M). The balance sheet is not distressed. The technical Graham F2 fail (total debt > working capital) is a capital-intensity artifact — OSAT fabs require continuous capex investment in bonding/test equipment. Comparable to shr-015 (financial sector caveats) in spirit, though less extreme.
All values in USD M (converted at ~31.36 TWD/USD).
| Year | Revenue | EBIT | Net Income | EPS (USD/ADR) | FCF |
|---|---|---|---|---|---|
| 2022 (peak) | 750 | 133 | 110 | $3.15 | 125 |
| 2023 | 681 | 81 | 63 | $1.80 | 113 |
| 2024 | 724 | 61 | 46 | $1.32 | 27 |
| 2025 (trough) | 763 | 28 | 18 | $0.50 | 5 |
| Q1 2026 (ann.) | 885 | — | 64 | $2.88 | — |
The EPS trajectory is unambiguous: peak ($3.15) → trough ($0.50) → early recovery ($2.88 annualized from Q1). FCF collapsed in 2024-2025 due to heavy capex investment during the trough (preparing for the memory upcycle). Revenue held relatively stable ($680–763M) throughout — the margin collapse was the story, not revenue loss.
Gross margin: 2025 full-year ~12% vs 2022 peak ~20%+. Q1 2026 gross margin recovered to ~13.8%.
| Year | Dividend | Yield at payment price |
|---|---|---|
| 2016 | $4.37 | — |
| 2017 | (gap — not in yfinance data) | — |
| 2018 | $1.20 (two payments) | — |
| 2019 | $0.77 | — |
| 2020 | $1.21 | — |
| 2021 | $1.57 | — |
| 2022 | $2.92 | — |
| 2023 | $1.50 | — |
| 2024 | $1.11 | — |
| 2025 | $0.82 | 1.2% at $66 |
Dividends are tied to earnings (variable payout policy). The dividend has tracked the earnings cycle downward. With FY2025 EPS of $0.50 and a $0.82 dividend, the payout ratio is 163% — ChipMOS is returning retained capital from prior peak years. FY2026 dividend (to be paid Jun/Jul 2027) will likely be set at $0.50–1.00 based on 2026 earnings. The "7%+ yield" in the thesis hook existed only when the stock was near $15; it has since been arbitraged away.
No insider transaction data available via Finnhub or yfinance for the ADR. ChipMOS is a Taiwan-incorporated company; insider disclosures are filed with TWSE (Taiwan Stock Exchange), not the SEC. The shr-002 insider signal cannot be applied here. This is a structural limitation for any Taiwan ADR in this framework.
Zero analyst coverage via Bloomberg/yfinance on the IMOS ADR. The last upgrade/downgrade recorded was Morgan Stanley in 2015. No consensus price target exists. Occasional coverage appears in Asian financial press and specialist semiconductor research (BofA Asia Tech conference March 2026), but no published ratings or targets were available. This creates two risks: (a) no valuation anchor and (b) no early-warning from analyst revisions.
| Quarter | Actual EPS | Est. EPS | Surprise |
|---|---|---|---|
| Q1 2026 | $0.72 | $0.782 | -8% (MISS) |
| Q4 2025 | $0.707 | $0.530 | +33% (BEAT) |
| Q3 2025 | $0.50 | $0.139 | +259% (BEAT) |
| Q2 2025 | -$0.75 | +$0.354 | -312% (MISS) |
The earnings surprise pattern reflects a rapid, volatile recovery from trough. Q2 2025 was the bottom. Three consecutive beats (Q3-Q4 2025, Q4 2025) drove the stock higher. Q1 2026 was a slight miss vs. elevated expectations — the first miss in three quarters at exactly the point where the stock hit an all-time high. This is a classic setup where re-rating momentum meets deceleration in surprise momentum.
Revenue Q1 2026: $216.4M, +25.4% YoY. Company stated demand exceeds capacity in memory/DRAM packaging.
The AI memory narrative is driving the re-rating. No structural dilution or financing concerns detected.
ChipMOS operates in two segments:
Display Driver IC (DDI) packaging (~60-65% historical revenue): Used in TV panels, monitors, smartphones. This segment was the core of the 2022 peak. It remains soft — Chinese TV/monitor panel OEM demand has not recovered to 2021-2022 levels. No near-term catalyst visible.
Memory/DRAM test & assembly (~35-40% revenue, growing): DRAM demand driven by AI servers, HBM packaging for NVIDIA/AMD. This is the Q1 2026 driver. Early-to-mid cycle for AI memory — real demand, not speculative.
Cycle position assessment: Q2 2025 was the trough. We are now in recovery/early upcycle. The stock has done 340% from trough to today. The question is how much further the recovery has to run and whether forward earnings justify the current multiple. At 23x annualized Q1 EPS, the market is pricing a full recovery and continued AI ramp. This is no longer a trough setup — it is a momentum/growth trade.
| Filter | Verdict | Key Number | Trap or Real Fail? |
|---|---|---|---|
| F1 Adequate size (>$2B) | PASS (marginal) | $2.30B mkt cap | Real pass — barely |
| F2 Financial strength (CR≥2, debt<WC) | FAIL | CR=2.44✓ / total debt>WC✗ | Cyclicality artifact (capex debt) |
| F3 Earnings stability (5y positive) | COND PASS | Full years positive; Q2 2025 loss quarter | Cyclicality artifact |
| F4 Dividend record (uninterrupted) | COND PASS | 9 years; 2017 gap; declining amounts | Semi-real: payout unsustainable |
| F5 Earnings growth (≥33%) | FAIL | EPS -84% (2022→2025) | Cyclicality artifact — but no growth |
| F6 Moderate P/E (≤15x) | FAIL | LTM P/E 56x; forward P/E 23x | VALUE TRAP signal at this price |
| F7 P/E × P/B ≤ 22.5 | FAIL | 56 × 2.93 = 164 vs 22.5 | VALUE TRAP signal at this price |
Score: 1 clean PASS, 2 conditional passes, 4 fails.
F6 and F7 failures are NOT cyclicality artifacts — they are direct consequences of the stock re-rating from $15 to $66. Graham's multiples tests fail because the valuation is no longer trough-priced. The thesis premise (trough valuation) has been invalidated by the price.
F2 and F5 failures are genuine cyclicality artifacts: the balance sheet reflects capex financing (not financial distress), and EPS fell due to industry trough (not structural impairment).
| Growth Rate | V (trailing $1.18) | MoS vs $66.05 | V (forward $2.88) | MoS vs $66.05 |
|---|---|---|---|---|
| g = 0% | $10.03 | -85% | $24.48 | -63% |
| g = 3% | $17.11 | -74% | $41.76 | -37% |
| g = 5% | $21.83 | -67% | $53.28 | -19% |
| g = 7.5% | $27.73 | -58% | $67.68 | +2% |
| g = 10% | $33.63 | -49% | $82.08 | +24% |
Break-even growth rate: 23.7%/yr on trailing EPS; 7.2%/yr on forward EPS.
The 7.2% break-even on forward EPS is achievable if the AI memory upcycle sustains. But "achievable" is not a margin of safety — Graham buys margin of safety, not break-even assumptions. The forward EPS itself ($2.88, Q1 annualized) may not hold: display IC remains soft, and memory cycles always overshoot and revert.
| Category | Signal | Assessment |
|---|---|---|
| Price action | RED | At 52-week high; +116% YTD; +16.4% today |
| Earnings/news | AMBER | Recovery real; Q1 miss vs elevated expectations |
| Insider direction | UNKNOWN | Taiwan domicile; no SEC Form 4 data |
| Analyst coverage | RED | Zero; no price target anchor |
| Dividend status | RED | 1.2% yield; payout 163%; next cut likely |
| Balance sheet | GREEN | Net debt $82M; current ratio 2.44x |
| Short interest | GREEN | 0.14%; not a squeeze/bear pressure |
| Dilution | GREEN | No ATM/converts/bond-for-equity found |
| Cycle timing | RED | Mid-recovery rally priced in; NOT trough entry |
4 RED, 1 AMBER, 3 GREEN, 1 UNKNOWN. The red flags are structural to the current price level, not company-specific problems that could resolve. The thesis hook was valid when the stock was at $15; at $66 it has been arbitraged away.
IMOS (Taiwan ADR, small-cap OSAT) is NOT in VWCE at material weight. Zero duplication with existing portfolio holdings. This structural advantage existed at the thesis hook price; it still exists today. But VWCE overlap is not the gating criterion — Graham valuation is.
RED — FAIL. The trough Graham setup has been fully priced. Do not buy at current levels.
The thesis hook was valid: ChipMOS is a clean balance sheet OSAT with a real AI memory tailwind, no VWCE overlap, and no dilution concerns. But the market discovered this before us. From the Q2 2025 trough at ~$15, the stock is up 340%. The four Graham valuation filters (F6, F7 most critically) now fail not as cyclicality artifacts but as genuine mis-pricing signals. At $66, the stock requires 7%+ annualized earnings growth just to break even on forward earnings. Graham's formula requires a margin of safety, not a break-even assumption.
WATCHLIST condition: If IMOS retraces to $28–35 (approximately 2.5–3.0x forward earnings at trough-recovery trajectory), re-run this screen. The underlying business quality and VWCE differentiation are real. The entry price just needs to reflect them.
Kill conditions that would remove even the watchlist entry: (1) Display IC segment shows structural demand destruction (OLED adoption eliminating DDI need); (2) memory packaging work shifts to Korean/Chinese OSATs at scale; (3) management initiates ATM or convertible financing; (4) dividend cut to zero.
VERDICT LINE: ChipMOS is a fundamentally sound OSAT with a real AI memory catalyst, but at $66 (52-week high, +116% YTD, +340% from trough) it fails four Graham filters outright — the trough value setup described in the thesis hook was real but has already been fully exploited by the market; no margin of safety remains at current price.