Rating: Cautiously Bullish Target Price: HKD 180–225 (base-case fair value range; probability-weighted expectation approx. HKD 196) Current Price: HKD 157.40 (close of 2026-09-02, price anchor for this report) ｜ Market Cap: HKD 66.0 billion (419.4 million shares) Margin of Safety: approx. +14% (based on the lower bound of base-case fair value at HKD 180) Time Horizon: 12 months ｜ Report Date: 2026-09-03
ASMPT is in a dual-delivery phase of orders and earnings driven by AI infrastructure buildout: 2026H1 new orders of HKD 12.75 billion (+85.1%), book-to-bill ratio of 1.43 (highest since 2021H1), Q2 revenue of HKD 4.94 billion (+52.1%) exceeding the top end of guidance, and adjusted net profit of HKD 973 million (+230.5%). However, the delivery structure is broader than "advanced packaging" alone—H1 advanced packaging revenue grew only +17%; the real drivers are AI server SMT, China mainstream wire bonders/die bonders, and photonics (roughly +2x). After a 131% rise over the past year and a 34% drawdown from the 52-week high of HKD 238.8, the current price of HKD 157.4 equates to only 24x J.P. Morgan's 2027 EPS forecast (HKD 6.56), below pure-play peer Besi's forward 29x and below the most bearish street target (Goldman Sachs at HKD 177). Based on explicit ramp modeling, we arrive at a base-case fair value of HKD 180–225 (+14% to +43% vs. current price), judging the stock to be modestly undervalued: growth options account for roughly 60% of the current price, requiring three catalysts—2026Q3 results (late October), HBM4 equipment wins, and the SMT segment strategic review—to validate in succession over the next two quarters. The main tail risks are an AI capex pullback in 2027 combined with HBM4 wins going to competitors (bear case HKD 60–75, roughly -52% to -62%) and export-control exposure given China's 42% revenue share.
One-liner: Mid-stage of the AI advanced packaging upcycle, with dual confirmation from orders (book-to-bill 1.43) and earnings (adjusted net profit +230.5%). Current price is approx. 23% below the consensus target (HKD 205.67) and still approx. 12% below the most conservative post-earnings sell-side anchor (Goldman Sachs at 177), presenting a deployable opportunity over the next 12–18 months—validation points are Q3 order momentum and HBM4 wins, while AI capex pullback and China export controls are the main falsification triggers.
Evidence Grade: B (the inflection is confirmed by both order and earnings indicators, but key data on HBM4 share and TCB market size diverge between company and third-party figures, and the memory-side win has not yet materialized). Environment Assessment: Tailwind.
Changes underway (three main threads):
Core drivers and leading indicators: ① Quarterly orders and book-to-bill (currently 1.43; company guides Q3 orders to grow high-single-digit QoQ—clear deceleration from Q2's +24.8%); ② HBM4 win announcements (historically the company has issued press releases for every major order); ③ Downstream AI capex chain (TSMC's October earnings call capex outcome and CoWoS expansion guidance; customer prepayments up +78.8% over six months to HKD 1.95 billion corroborate order authenticity).
Expectation Gap (mixed): The market anchor has been verified—6 sell-side analysts covering, consensus target of HKD 205.67, zero sell ratings; the current price is 23% below that anchor. The market is currently pricing in "moderate growth" (reverse-engineering the current price implies a perpetual growth rate of approx. 6.6%), while the order book and CoWoS/HBM demand math support a higher 2026–2028 earnings path; but if AI capex pulls back in 2027, the crowded consensus-bullish structure will amplify a downward-revision chain. Four-dimension assessment: change 3/3, drivers 3/3, expectation gap 2/3, catalysts 3/3.
The surge in orders and revenue is fact, and the sharp rise in prepayments provides partial deposit backing for the order book. But two structural truths must be noted: first, advanced packaging's +17% growth is less than half the Group's +42.5%; the real engines of this round are AI server SMT (segment orders hit a record HKD 3.73 billion), China mainstream wire bonders/die bonders (China revenue share rising to 42.2%), and photonics (pluggable optical module revenue approx. USD 75 million, nearly 3x YoY)—"AI demand" is multi-faceted, and TCB advanced packaging is still catching up on the revenue line. Second, official signals of a flattening momentum slope have emerged: the company's Q3 order guidance of "high-single-digit QoQ growth" is an order of magnitude below Q2's +24.8% QoQ, and Goldman Sachs cut its target from 206 to 177 in early August precisely on this basis. The "high-slope phase" of the order delivery period has most likely passed in 2026H1; 2027 order growth will naturally decelerate against the higher base—this defines the validation window for this thesis as 2026Q3–Q4.
The demand-side chain is quantifiable: CoWoS capacity more than doubling over two years, HBM stacking evolving from 8 to 12/16 layers nearly doubling bonding steps per package, and optical modules plus CPO opening a second curve—this is an exogenous platform shift, not organic industry growth. But the market size figures must be presented in parallel: the company's 2028 TCB TAM of over USD 1.6 billion (implying approx. 40% CAGR 2025–2028) is 45–70% higher than third-party Yole's 2030 estimate of USD 900 million–1.1 billion—the company has an incentive to inflate the TAM to support its re-rating narrative. Share reality: leadership in logic-side C2S TCB (50+ tool repeat major order), but the announcement only claims "consolidating leadership" with no mention of "exclusivity"; memory-side HBM4 wins are not locked (Korea revenue -64% is the evidence), and the company has only signed an "exclusive joint evaluation program" with one major memory maker (evaluation is not a win). Our base case requires a 2025–2028 revenue CAGR of approx. 20–26%: even on Yole's conservative TAM figures this growth rate is broadly supportable, but the upper half of the company's "35–40% share target" depends on memory wins not yet secured—this is why the thesis is downgraded from 0.70 to 0.60.
The 2026H1 earnings improvement is real: adjusted gross margin 41.2% (Q2 42.4%), adjusted net margin 10.9%, OCF/net profit repaired to 1.25, net cash of HKD 3.63 billion. But three disciplines must be applied: ① the tax rate jump from 1.6% to 32% is structural (German business turning profitable and taxed at the local blended rate + global Pillar Two minimum tax), not one-off—all EPS forecasts must be normalized at a 29–32% tax rate; ② FY2025's HKD 902 million net profit was flattered by the disposal gain; core profit was approx. HKD 500 million after excluding one-offs; the company's self-defined "adjusted profit" differs by approx. HKD 96 million across two official documents without explanation, impairing cross-period comparability—hence this report uses "Q2 single-quarter adjusted profit of HKD 638 million × 4 = annualized approx. HKD 2.55 billion (EPS HKD 6.08)" as the run-rate floor—and notes it includes the portfolio peak of SMT's 36.8% quarterly gross margin (conservative estimate approx. HKD 5.2); ③ the working-capital strain problem of FY2025's full-year OCF of only HKD 219 million (FCF turned negative) was repaired in 2026H1, but inventory (+15%) and receivables (+28.5%) are still expanding, requiring continued monitoring of cash conversion.
An honest statement of the valuation judgment: the current price is not deeply undervalued, but the growth expectations priced in are modest. Three pieces of evidence: ① the street's most conservative 2027 EPS (J.P. Morgan 6.56) on the purest peer Besi's forward 29.2x yields a fair value of approx. HKD 192, still 20% above the current price—i.e., the "cheapness" does not depend on our being more optimistic than the street on earnings; ② PB of 3.81x sits at the 93rd percentile of the past 5 years, indicating the structural re-rating has largely occurred with limited room for further multiple expansion—upside must come mainly from earnings upgrades; ③ 6 covering analysts with zero sell ratings and a consensus target implying +31% is itself a crowding signal—if Q3 orders land at the low end of guidance or HBM4 wins go to Korean competitors, consensus downgrades will resonate with multiple compression. Base-case fair value is HKD 180–225 (+14% to +43%), probability-weighted expectation approx. HKD 196 (+25%); the bear case of HKD 60–75 (-52% to -62%) is a tail that must be squarely faced.
| Metric | FY2023 | FY2024 | FY2025 | 2026H1 (as of 2026-06-30) |
|---|---|---|---|---|
| Revenue | 14,697.5 | 13,230.0 (restated continuing ops 12,483.2) | 14,521.0 (incl. discontinued) / 13,736.2 (continuing, +10.0%) | 8,902.6 (+42.5%) |
| Net profit attributable to shareholders | ≈711* | 342.0 (-51.9%) | 901.9 (+163.6%; continuing 1,084.7) | 589.0 (+174.1%) |
| Adjusted net profit (non-HKFRS) | — | 426.0 (incl. NEXX) | 466.7 (continuing) / 501.4 (incl. discontinued; interim restated 597.6**) | 972.7 (+230.5%) |
| Gross margin | 39.3% | 40.0% | 37.8% (continuing) | 41.1% (adjusted 41.2%; Q2 42.4%) |
| Adjusted net margin | — | 3.2% | 3.4% | 10.9% |
| Operating cash flow | 2,128*** | 945*** | 219*** | 727.3 |
| Free cash flow | ≈1,786 | ≈618 | -137 | 590.5 |
| Cash + deposits over 3 months | — | — | 5,675.6 (2025-12-31) | 5,875.9 (2026-06-30) |
| Interest-bearing debt (bank borrowings) | — | — | 2,393.8 (2025-12-31) | 2,250.0 (2026-06-30) |
| Net cash | — | — | 3,282 | 3,625.9 |
| Gearing ratio | — | — | — | ≈18% (estimated) |
| New orders | ≈12,260 (estimated) | 12,750 (+4.0%) | 14,477.8 (continuing) | 12,753.8 (+85.1%) |
| Book-to-bill | ≈0.83 | 0.96 | 1.05 | 1.43 |
| DPS (HKD, incl. special) | — | 0.67 | 1.39 | Interim 0.97 (+273%) |
* Backed out from "-51.9%" in the FY2024 results announcement. ** The two adjusted-profit versions differ by approx. HKD 96 million without company explanation; both are shown. *** FY2023–2025 full-year cash flows are third-party database anchored values (annual report originals not in this document set, pending verification); 2026H1 figures are from the interim report.
Drivers of metric changes (extract of YoY ≥±20%, all per management commentary): FY2024 revenue -10.0%—consumer electronics weakness, soft auto/industrial end markets, slow non-AI semiconductor recovery; FY2025 net profit +163.6%—dense non-recurring items (JV disposal gain +1.11 billion, restructuring -340 million, impairments -280 million, inventory write-off -70 million); 2026H1 revenue +42.5%—AI infrastructure expansion, SMT +53.1% (AI servers + China EV), SEMI +35.3%; Korea revenue -64.2%—memory customers' new equipment purchase decisions affected by the HBM4 launch timeline; China revenue +57.7% (share 38.1%→42.2%)—wire bonders/die bonders, AI capacity expansion and high OSAT utilization; income tax HKD 4 million→349 million—German tax + Pillar Two + prior-year under-provision; customer prepayments +78.8%—not specifically explained by the company, mutually corroborated by record order backlog.
Q2 revenue of HKD 4.936 billion (USD 630 million, +52.1% YoY / +24.4% QoQ), above the top end of company guidance (USD 540–600 million); Q2 adjusted gross margin 42.4% (+290bps QoQ), SMT adjusted gross margin 36.8% (+429bps, driven by product mix and volume). H1 adjusted net profit of HKD 973 million (+230.5%), of which Q2 alone was HKD 638 million. Orders: Q2 new orders of HKD 7.08 billion (+97.6%), SMT segment orders at a record HKD 3.73 billion, SEMI at USD 428 million (+39% QoQ, driven by wire bonders/die bonders and photonics). Interim dividend of HKD 0.97 (+273%). Beat or miss: Citi noted Q2 adjusted net profit beat its forecast by 76% and book-to-bill at a five-year high; on the day after results the stock rose as much as 15% intraday, with BofA/Citi/J.P. Morgan setting targets of 310/250/225 respectively—the good news was quickly priced in. Forward guidance: Q3 revenue of USD 630–690 million (midpoint +46.3%, company calling it "above market expectations"), Q3 orders growing high-single-digit QoQ (momentum downshift). Since August, amid the sector pullback (SOX -6% in a single day in July, another slump mid-August), the stock has drawn down 34% from its 238.8 high—the market is digesting the combination of "decelerating order guidance + crowded AI valuations," which is precisely the source of today's odds.
Business model snapshot: Order-driven equipment manufacturing, asset-light (capex/revenue around 2.5–3%), no recurring revenue disclosed — revenue is driven by individual equipment orders, with two segments (SMT and SEMI) covering three demand categories: "AI front-end installation (advanced packaging) + AI spillover (server assembly/power/optical modules) + replacement demand (wire/die bonding)." Pricing power comes from POR (process of record) lock-in: once certified, customers are deeply bound; the 35–40% TCB target market share and C2S process position are the sources of the technology premium; however, the Group's gross margin (41.2%) is significantly below pure advanced packaging peers (Besi 63%, Disco 71%), due to the low-margin SMT business and the high share of mainstream (non-advanced) equipment in China.
Cash content of earnings test: OCF/net profit for FY2023–2026H1 was 3.28 → 2.76 → 0.24 → 1.25; FCF/net profit was 2.76 → 1.81 → -0.15 → 1.01. FY2025 is the anomaly: working capital outlays of HK$950 million in inventory and HK$1.01 billion in receivables turned FCF negative, and the company did not specifically explain this in its disclosures — this is a cross-period anomaly to monitor, not noise to be ignored. 2026H1 showed recovery (OCF HK$727 million), but working capital again absorbed approximately HK$2.14 billion in the half year (inventory HK$7.25 billion/+15%, receivables HK$4.56 billion/+28.5%); fortunately, customer prepayments of +HK$860 million provided an offset, and the receivables/annualized revenue ratio (25.6%) was flat versus the prior year-end. Recurring earnings test: FY2025 net profit attributable to shareholders was HK$902 million vs. approximately HK$495 million after excluding one-offs — half of book profit came from disposal gains; core earnings trend (HK$467 million adjusted, +9.6%) is far smoother than the headline +163.6%; reverse-checking the company's own adjusted figures: add-backs are mainly SBC (HK$71 million in H1, roughly 7% of adjusted net profit), which is moderate, but the basis drifted by HK$96 million between two documents — weak discipline.
Return on capital: ROIC FY2025 approximately 3.8% (cyclical trough) → 2026H1 annualized approximately 13.6%, now crossing WACC (approximately 11%) — this report judges that its durability depends on order persistence, not on an entrenched moat premium. Maintenance capex: CapEx/depreciation 0.41–0.57 (FY2023–2026H1), <1, a cash-cow profile — cash generated does not need to be reinvested at scale to maintain capacity. Red flag list: ① Earnings highly dependent on the single AI engine (advanced packaging is 30% of revenue; management explicitly stated automotive/industrial end-markets remain weak in the near term); ② Sharp drop in FY2025 cash conversion (OCF/NP 0.24); ③ Highly volatile order cycles (B2B 0.83→0.96→1.05→1.43). Top five customers account for only 19% of revenue — healthy concentration.
Consistency of words and deeds: pragmatic, conservative guidance, high delivery rate. The last three quarters' revenue guidance was all met or exceeded the upper bound: 2025Q1 guidance US$370–430 million (actual approximately US$400 million); 2026Q1 guidance US$470–530 million (actual US$508 million); 2026Q2 guidance US$540–600 million (actual US$630 million, above the upper bound). The 2025 capex plan of approximately HK$350 million vs. actual HK$356 million — delivered. "Double TCB capacity by end-2025" has no direct documentary verification; recorded as partially delivered.
Shareholder friendliness: generous dividends, no buybacks; verdict "shareholder-friendly (dividend-type)". FY2024 full-year dividend of HK$0.67 (including special HK$0.25), 82% of net profit attributable to shareholders; FY2025 dividend of HK$1.39 (including special HK$0.79, linked to the one-off gain on the JV disposal), 64% of net profit attributable and 124% of adjusted; 2026 interim dividend of HK$0.97 (+273%), approximately 50% of adjusted. No buybacks in 2026H1; the ESAS employee share award scheme increased share capital by +0.7% since the beginning of 2025 — mild dilution.
Risk signals: ① CEO transition — Robin Ng announced his intention to step down in February 2026 and formally handed over to Bassel Haddad on August 11 (14 years at Intel, former Senior VP at SkyWater; annual salary approximately S$905,000 plus a S$1 million joining bonus); the company confirmed no disagreements with the board — an orderly transition, but his stance on the SMT disposal and R&D resource allocation needs to be verified at the Q3 results call in late October; ② Major shareholder ASM International holds a stable 24.55% with two non-executive directors on the board, but there are precedents of a stake reduction in 2017 and the termination of privatization discussions in 2024 — a potential supply-shock tool; ③ In 2026 to date: no controlling shareholder stake reduction, no open-market sales by directors/supervisors/senior management, no pledge disclosures.
| Segment | 2026H1 Revenue (HK$100mn) | Share | Adjusted Gross Margin | YoY | Business Logic in One Sentence |
|---|---|---|---|---|---|
| Semiconductor Solutions (SEMI) | 50.36 | 56.6% | 46.5% | +35.3% | TCB/flip chip/photonics/wire bonding equipment; POR lock-in + technology premium; the main battlefield for AI advanced packaging |
| Surface Mount Technology (SMT) | 38.67 | 43.4% | 34.2% | +53.1% | PCB assembly equipment; AI server motherboards + China EVs drive demand; intense competition, high earnings elasticity |
SEMI is the profit engine: Based on H1 segment profitability estimates (SEMI segment margin approximately 19.9%, SMT approximately 11.0%), SEMI contributed approximately HK$1.00 billion of segment profit vs. approximately HK$430 million for SMT — about 70% of segment profit comes from SEMI, which is 56.6% of revenue, and its margin (20.9% in Q2) is still climbing. The gross margin gap between the two segments is 12.3 percentage points (≥10pp requires explanation): SEMI contains high-margin products like TCB/photonics and process lock-in premiums; SMT is relatively standardized assembly equipment facing direct competition from Chinese and international peers, with gross margin fluctuating widely with product mix (36.8% in Q2 alone was the mix peak, in sharp contrast to the trough in 2025H1 when segment margin was only about 1.9%). SMT is under strategic review (launched 2026-01-21): both orders and profitability are currently high (record orders of HK$3.73 billion, 11.0% margin), which both raises the asking price for a sale and may reduce willingness to sell — this report's base case excludes divestment premium; the SMT conclusion is treated as a two-way option.
Accounting red flags (pattern + severity + documentary evidence):
Cross-period consistency: ① OCF/net profit attributable 3.28→2.76→0.24→1.25 — the FY2025 plunge was unexplained by the company (the 2026H1 cash flow statement showing a net operating cash outflow of HK$113 million in 1H25 corroborates the trough); verdict "unexplained by company"; ② Effective tax rate 1.6%→approximately 17%→32.0%, with "under-provisioning in prior years" recognized in two consecutive periods (HK$18 million/HK$61 million); management's explanation (German tax + Pillar Two) is consistent with the notes; ③ Adjusted profit basis drift as above, unexplained by company. Overall verdict: two medium-severity discipline issues in the reporting basis; no evidence of fraud-level misconduct; this report uses a dual-track comparison of "statutory + self-tested exclusion of one-offs" for earnings throughout, and does not solely rely on the company's adjusted figures.
Current market data and multiple percentiles (price anchor: HK$157.40, 2026-09-02 close): market cap HK$66 billion; PE(TTM) statutory 51.7x (24% over the past 3 years / 56% over 5 years percentile — TTM earnings include large one-off items and are in a steep inflection phase, percentiles are for reference only, not the headline on cheapness); PE(TTM) adjusted 57.6x; forward PE: FY2026E approximately 30.8x (self-tested adjusted EPS midpoint 5.12) / 37.1x (BofA statutory HK$4.24 basis), FY2027E 19.5–24.0x (self-tested 8.08 / J.P. Morgan 6.56); PEG approximately 0.7; EV/Sales (FY2026E) 3.17x; PB 3.81x (93% percentile over 5 years — structural re-rating is largely done, must be viewed alongside ROE uptrend delivery); dividend yield approximately 0.9% (FY2025 DPS basis). Average daily turnover approximately HK$480 million (approximately 0.73% of market cap) — normal liquidity.
Peer valuation benchmarking (exit multiple anchor):
Note: Hanmi's TTM of approximately 98x represents the ceiling of scarcity premium for a single major customer, not a benchmark anchor; Besi's forward multiple shows a basis split of 29–46x across data sources (NTM vs CY2027); this table uses the Yahoo basis and flags the uncertainty.
What expectations does the market already imply: The current price of 157.4 = J.P. Morgan 2027 EPS 6.56 × 24.0x, or 25.9x the current Q2 annualized run-rate EPS of 6.08; solving backward with net cash of HK$8.65/share deducted and WACC at 11%, the implied perpetual growth rate is only about 6.6%. In one sentence: the current price only requires the company to sustain the current run-rate and grow moderately over the long term; whereas the order book (B2B 1.43, backlog of approximately HK$10 billion) and demand math (CoWoS +69%, HBM layers doubling, photonics 3x) support a path of roughly 30% earnings CAGR for 2026–2028 — there is a positive gap between implied expectations and actual earnings power, and the base case requires a 2025–2028 revenue CAGR of approximately 20–26%, not breached even under Yole's conservative TAM basis. Reverse check: if AI capex falls back in 2027, the company's FY2024 precedent (revenue -10%, adjusted net profit -42.8%) shows its cyclical elasticity is enough to knock EPS back below half the run-rate — this is precisely the source of magnitude in the bear case.
Three layers of value (EPV): Asset value (floor) HK$41.3/share (including A-share holdings fair value revaluation upside: book HK$2.15 billion vs. fair value HK$4.71 billion); EPV zero growth HK$63.9/share (normalized EPS 6.08 ÷ WACC 11% + net cash 8.65; conservative basis of 5.18 corresponds to approximately HK$55–58); growth option is approximately 59% of the current price. Nearly 60% of the current price is supported by the growth option — for a company that has hit the demand inflection point, this does not constitute evidence of overvaluation; the verdict rests with the ramp scenario; but it is the structural explanation for "why this stock falls 40–60% when capex expectations reverse."
Three scenarios and odds (unified method: 2028 adjusted EPS × exit multiple, discounted 2 years at WACC 11%):
| Scenario | Probability | Fair Range | vs. Current Price | Core Assumptions |
|---|---|---|---|---|
| Bear | 25% | 60–75 | -52% to -62% | 2027 AI capex downturn + HBM4 mandate lost: 2028 adjusted EPS 4.8–5.1 × 15–18x (benchmarked against the 2024 cycle-bottom PB≈1.5x and the downside of its own 10-year median PE of 25.7x); also includes China-related export control escalation as a low-correlation independent tail |
| Base | 50% | 180–225 | +14% to +43% | Ramp: 2026 revenue approximately HK$19.7 billion (+43%) → 2027 approximately HK$23.5 billion (+19%) → 2028 approximately HK$27.5 billion (+17%); 2028 adjusted EPS 8.4–9.2 (blend of J.P. Morgan path 8.07 and self-tested 9.18) × 26–30x (anchors: Besi forward 29.2 / K&S approximately 24 / Disco 42.2; own 10-year median 25.7 as floor reference); equivalent to 2027 EPS≈7.3 × 28x |
| Bull | 25% | 295–330 | +87% to +110% | TCB share 35–40% achieved + HBM4 fluxless qualification success + hybrid bonding Gen-2 converting to orders: 2028 EPS≈11.1 × 33–37x (between Disco 42.2 and the upper bound of Besi's recovery) |
Bear case design note: The most pessimistic published forecast is Goldman Sachs' 2026-08-03 Neutral target price of HK$177 (pricing only "order deceleration," implying no earnings decline); this report's bear case of 60–75 is significantly below that anchor, additionally pricing the combination of "AI capex downturn + HBM4 failure" (2022–24 cycle precedent: revenue -10%, adjusted net profit -43% supports the elasticity). The base case valuation implication (202.5 midpoint) falls between Goldman's 177 and BofA's 310, broadly in line with the consensus target price of 205.67 — we do not claim the market is systematically wrong, only that the current price has not yet caught up with consensus.
In-house earnings forecast (results reconciliation anchor):
| Period | Revenue (HK$100mn) | Adjusted Net Profit (HK$100mn / EPS) | Benchmark |
|---|---|---|---|
| FY2026E | 197 (190–204, continuing operations, +38% to +48%) | 21.5 (19.7–23.2; EPS 4.70–5.53) | Management's Q3 guidance of US$630–690 million is embedded; above J.P. Morgan's revenue of HK$18.4 billion (forecast formed before Q2) |
| FY2027E | 237 (228–246, +17% to +26%) | 34 (31.1–36.8; EPS 7.40–8.76) | J.P. Morgan's 2027 EPS of 6.56 is below the lower bound of our range; the difference lies mainly in net margin assumptions (our 13.5–15% vs. their implied approximately 12%) |
Conclusion: modestly undervalued, target price HK$180–225 (base case fair value), margin of safety approximately +14% (base case floor basis); probability-weighted expectation approximately HK$196 (+25%). Evaluate quality (order delivery + earnings ramp + net cash) and price (93rd percentile PB + crowded consensus) separately: a good company, an acceptable price, and profound tails — odds lean positive but not lavish; the pace of catalyst delivery will determine success or failure.
Industry size. ASMPT operates in semiconductor back-end equipment (assembly/packaging plus test equipment), overlaid with an SMT placement equipment business. Scale by layer: global semiconductor equipment total sales hit a record USD 135.1 billion in 2025 (+15%); SEMI forecasts USD 165.9 billion in 2026 (+23.2%) and USD 229.5 billion in 2028 (five consecutive years of growth). Within that, assembly & packaging (A&P) equipment goes from ~USD 6.1 billion in 2025 → ~6.7 billion in 2026F → 8.6 billion in 2028F (SEMI, 2026-07, primary source); Yole's back-end equipment scope (bump/dicing/thinning/metrology & inspection) goes from USD 6.9 billion in 2025 → 9.8 billion in 2030, a 7.1% CAGR. Structural bright spots far outpace the overall market: TCB equipment at ~USD 0.9–1.1 billion by 2030 per Yole (13–14% CAGR), hybrid bonding at 24.6% CAGR, while the company's own figure puts the 2028 TCB TAM above USD 1.6 billion (implied ~40% CAGR) — as noted in C2, the company's numbers run 45–70% above third-party estimates, and this report uses both sets as upper and lower bounds. Advanced packaging devices (Yole, cited pending verification): ~USD 46 billion in 2024 → ~79 billion in 2030.
Quantified chain of the demand inflection (this report's judgment: the exogenous structural inflection is confirmed): driver trajectories — TSMC CoWoS monthly capacity of ~70–80k wafers in 2025 → 120–140k in 2026 (~+69%, TrendForce), with third-party estimates by end-2027 revised up further to 200–220k wafers/month (2.7–2.9x over two years); SK hynix plans to expand 1c-node monthly capacity from ~20k to 160–190k wafers by end-2026. Unit consumption — HBM stacking of 8 layers (HBM3) → 12 layers (HBM3E/HBM4) → 16 layers (in progress for 2027), with each added layer meaning one more wafer-level bonding step; 16 layers doubles the bonding passes versus 8. On the logic side, composite dies are evolving toward reticle-level mega-chips, raising the intensity of C2S TCB tooling per wafer in tandem. Penetration — HBM stack bonding is already 100% dependent on TCB-type processes; hybrid bonding is unlikely to enter before HBM4e/16 layers (~2027–28) (TrendForce 2026-07; SK hynix confirmed at Hot Chips 2026 that HBM4E will not use hybrid bonding, deferring it to HBM5 — the TCB golden window is being extended). Company share — leadership in logic C2S (50+ unit repeat large order); on the memory side, starting from a ~5% installed share in HBM3/3E and entering SK hynix HBM4 qualification via fluxless TCB (THE ELEC: about half of hynix's current ~50 HBM4 TCB units were supplied by ASMPT). Photonics is the second curve: pluggable optical module solution revenue of ~USD 75 million in H1 (~3x), CPO market from ~USD 330 million in 2026 → ~3.37 billion in 2034 (Fortune Business Insights).
Value chain and value distribution. Upstream: custom components such as precision motion platforms, vision systems and thermocompression heads; equipment makers hold medium bargaining power (the company noted extended lead times for some materials, occasional bottlenecks). Midstream is the equipment makers themselves. Downstream: three buyer types — foundries with in-house advanced packaging (TSMC CoWoS/SoIC, the most aggressive expanders), memory IDMs (SK hynix/Samsung/Micron HBM packaging lines), and OSATs (ASE/Amkor/JCET/PTI) taking outsourced work. Gross margin distribution confirms "equipment layer eats the meat, packaging/test layer drinks the soup": Disco gross margin 71.3%, Besi 62–63%, K&S 47.8%, ASMPT group 41.2%, while at the OSAT layer Amkor is only 14% and ASE's ATM segment 20–27% — the incremental value of advanced packaging accrues mainly to equipment and foundry. ASMPT's POR process lock-in gives it strong downstream bargaining power, but group gross margin is diluted by low-margin SMT and China mainstream equipment.
Supply-demand and competitive landscape. Advanced packaging is tight-balanced to short: CoWoS undersupplied through end-2026 (gap 20% → 10%); HBM still structurally short in 2026 (Micron's 2026 HBM capacity already sold out); mainstream packaging is seeing a modest consumer/industrial/auto recovery, with high China OSAT utilization driving wire bonder/die bonder demand. Competitive structure: back-end equipment overall is "one dominant, several strong" — Disco leads in dicing/thinning (~60–70% global share, secondary source pending verification), with ASMPT/Besi/K&S/Hanmi in the leading group. By sub-segment: HBM TCB is dominated by Hanmi + Hanwha + ASMPT (SK hynix HBM3/3E installed share ~45%/45%/5%, channel research pending verification); ASMPT leads logic C2S TCB; hybrid bonding is led by Besi + Applied Materials; wire bonding is a duopoly of K&S and ASMPT. Entry barriers are extremely high: sub-micron bonding-precision know-how, POR qualification lock-in (high switching costs), 2–3 year joint development cycles with leading foundry/memory makers, and export-control compliance credentials. Substitution risk is about technology routes rather than price: hybrid bonding (pitch <5µm) could in the long run partially replace TCB in 16-layer-plus HBM — at the current point we judge this as window deferral, not acceleration; price competition and domestic substitution from Chinese local players (Xinyichang etc.) in mainstream die bonding is a long-term erosion factor.
Cyclicality and regulation. The industry is mid-way through an AI-driven upcycle: SEMI projects five consecutive years of total equipment growth through 2028; back-end equipment revenue +6.2%/+3.8% QoQ in 2026Q1/Q2 (Yole monitor). Leading indicators in order: book-to-bill (ASMPT 1.43), TSMC CoWoS capacity and gap, HBM layer roadmap and HBM4e qualification, SEMI test equipment shipments (2025 +55.3%), OSAT utilization. Regulation cuts both ways: US export controls added HBM to ECCN 3A090.c in December 2024 and cumulatively listed 160+ Chinese entities, but back-end packaging equipment is not currently a priority controlled category, and ASMPT is a Singapore-headquartered non-US entity — yet Mainland China contributed 42% of revenue in 2026H1 (mainly wire/die bonders), a policy-sensitive exposure; over the medium-to-long term, domestic substitution and China's equipment self-sufficiency will erode foreign players' mainstream-segment share (short-term resonance with AI expansion, long-term seesaw).
Peer operating benchmark (this company's data merged into the same table):
| Company | Revenue scale (latest period) | Revenue growth | Gross margin | ROE | Market/size position | Key differences vs ASMPT |
|---|---|---|---|---|---|---|
| ASMPT | FY2025 USD 1.76 billion (continuing operations) / 2026H1 annualized ~USD 2.27 billion | 2026H1 +42.5% | 41.2% (adjusted) | 2026E ~12–15% (upward inflection) | No.2 in back-end revenue per Yole (behind only Disco) | Widest platform: TCB + die bond + wire bond + photonics + SMT; captures both waves of AI demand |
| Besi | FY2025 EUR 591 million | FY2025 -2.7%; 2026H1 orders +116% | 62–63% | Not available | Global leader in hybrid bonding | Pure assembly equipment, betting on hybrid bonding; scale ~2/3 of ASMPT |
| K&S (KLIC) | Q3 FY26 quarterly USD 330 million | +122.6% (low base) | 47.8% | ~6.5% for the quarter | Global wire bonding leader | Strong wire, weak memory TCB (lacks HBM experience); higher China exposure at 56% |
| Hanmi (042700.KS) | FY2025 ~USD 420 million | FY2025 +3.2%; strong 2026Q2 | Operating margin 43.6% (gross margin not disclosed) | Not available | SK hynix's main HBM TCB supplier | Single large-customer dependence, narrow product line; market cap/revenue ratio ~40x prices in scarcity premium |
| Disco (6146.T) | Q1 FY3/26 annualized ~USD 3.1 billion | +27.1% | 71.3% | Not available | Dicing/thinning monopoly (60–70%) | Single-process monopolist; the purest HBM amplifier |
| Xinyichang (688383.SH) | 2025 RMB 727 million | -22.2% | Not available | Not available | China's local die bonder leader | Direct domestic-substitution rival; scale only ~0.8% of ASMPT, still a technology generation behind |
Company positioning and share-scope notes: ASMPT is one of the global top-tier duo in back-end assembly equipment (second only to Disco in revenue per Yole) — "mainstream-equipment duopolist + logic-side TCB leader, memory-side fast follower", with a rising share trend. Two scopes must be juxtaposed: the company claims "TCB leadership, targeting 35–40% share by 2028"; third-party channel research shows its HBM memory-side installed share at only ~5% (Hanmi ~45%, low reliability pending verification) — the gap stems from conflating "logic-side leadership" with "memory-side chasing"; investors should read the TCB narrative as "logic-side proven, memory-side an option". Moat sources: POR lock-in + the industry's broadest product portfolio (able to capture both AI front-loading and AI spillover demand) + depth of joint development with the leading ecosystem; weaknesses: group gross margin below pure-play peers, 42% of China revenue hanging on export policy, and an SMT discount.
Back-end equipment is an order-driven business; the quarterly book-to-bill ratio is the earliest leading indicator of revenue and cash flow, ahead of the P&L by 2–4 quarters. ASMPT's series:
| Period | New orders (HKD 100M) | YoY | Revenue (HKD 100M) | B2B | Backlog, period-end (HKD 100M) |
|---|---|---|---|---|---|
| FY2023 | ≈122.6 (estimated) | — | 147.0 | ≈0.83 | — |
| FY2024 | 127.5 | +4.0% | 132.3 | 0.96 | 60.5 |
| FY2025 | 144.8 (continuing operations) | +13.6% (+17.5% incl. discontinued scope) | 137.4/145.2 | 1.05 | 61.7 |
| 2026H1 | 127.5 | +85.1% | 89.0 | 1.43 (both Q1 and Q2 at 1.43) | ≈100 (estimated from opening 61.7 + orders 127.5 − revenue 89.0) |
Three readings: ① Current backlog of ~HKD 10.0 billion ≈ 5.3 months of revenue coverage, providing high visibility for 2026H2–2027H1 revenue — the foundation of this report's base case of +43% revenue in 2026; ② B2B took three years to go from 0.83 → 1.43; after the 2021 peak (>1.4) came a sub-1 digestion period through 2022–2024 — the historical amplitude warns that 1.43 cannot be extrapolated linearly, and the Q3 order guidance of high-single-digit QoQ growth (implying ~USD 950–990 million) is already a first signal of deceleration; ③ Customer prepayments of HKD 1.95 billion (+78.8%) corroborate order authenticity as deposits, but they cover only part of the order book (the company discloses that customers generally use credit + deposit terms), and cannot be extrapolated as default protection for all orders. By segment: Q2 SEMI orders B2B 1.16, SMT estimated ~1.82 — SMT orders are more volatile at highs (the capex nature of AI server placement), and in the next downturn SMT will fall first.
Judgment: cautiously bullish, confidence 0.60, time horizon 12 months. The current price of HKD 157.4 is ~14% below the lower bound of the base fair-value range of HKD 180 and ~29% below the range midpoint, with a probability-weighted expectation of ~HKD 196 (+25%); the odds structure is positive but not lavish — the tail depth of the bear case at 60–75 (25% probability) calls for position sizing rather than valuation management. Strategically: the current price offers allocation value; Q3 results (end-October) and the HBM4 socket award are validation points for adding; re-assess if the price breaks above 225 (safety margin exhausted) or below 120 accompanied by order deterioration. Quality upside (gross margin 41% → 42%+, net margin 11% → 14% direction) coexists with valuation constraints (PB at the 93rd percentile, zero sell-side consensus); earnings upgrades materializing are the only reliable upside engine.
Key risk disclosures: ① AI capex rolling over in 2027 — revenue and profit double-hit under an order air pocket (2022–24 precedent: revenue -10%, adjusted net profit -43%); B2B below 1.0 is the earliest signal; ② escalation of China export controls — China revenue at 42.2% and still rising; controls extending to back-end equipment or major customers being listed is a low-probability, high-impact tail; ③ HBM4 socket going to Korean players (Hanmi already won hynix's first order of USD 287 million; ASMPT's joint evaluation not yet converted); ④ erosion of the EPS anchor from the 32% structural tax rate and adjusted-scope drift; ⑤ hybrid bonding substitution timing (Samsung has begun building an HB production line; SK hynix deferred to HBM5 — window temporarily deferred but requires ongoing monitoring); ⑥ if the SMT strategic review ends in retention or a low-price outcome, the valuation framework switch fails.
Catalyst tracking calendar:
Core verification checklist: Q3 order absolute value and B2B (vs. high-single-digit QoQ guidance); TCB/photonics share of new orders and tone upgrades; HBM4 volume order press releases; transactional announcements on the SMT review; effective tax rate and cash conversion (OCF/net profit, inventory and receivables turnover); customer prepayments QoQ. Invalidation discipline: any of — B2B below 1.0, backlog declining QoQ, capex cuts by TSMC/major memory makers, BIS adding back-end equipment to China controls, full-year effective tax rate persistently above 28% with further scope drift — triggers re-reading of this report's bear case.
Data and scope notes: financial data per company announcements (FY = fiscal year ending December 31); 2026H1 balance-sheet stock as of 2026-06-30; FY2023–2025 full-year cash flows are third-party database anchored values pending verification; price anchor is the 2026-09-02 close of HKD 157.40 (frozen for this run); this report was generated by Loyan AI and does not constitute investment advice.