Layer 4 · 522 — ASMPT Limited
One-line thesis
ASMPT is the TC bonder share-taker that just won the SK Hynix HBM4 socket from Hanmi with bookings +72% YoY — own it as the back-end equipment leader at the inflection point in HBM stacking economics, with hybrid bonding and HBM5 as the next-cycle optionality and a Hong Kong listing as the geographically-distinctive vehicle.
What ASMPT physically does
ASMPT (formerly ASM Pacific Technology, Hong Kong-listed) is the global #2 semiconductor back-end equipment vendor after Hanmi for thermo-compression bonding (TC bonding) — the precision pick-and-place machines that stack HBM dies vertically with sub-micron alignment under controlled heat and pressure. Across the company's full portfolio, ASMPT also makes die-attach, wire bonders, mold-equipment for advanced packaging, and SMT (surface-mount-technology) lines for downstream assembly. But the AI-stack-relevant business is the TC bonder unit and increasingly the adjacent hybrid-bonding roadmap.
In an HBM line at SK Hynix, Micron or Samsung, the TC bonder is the single most expensive piece of equipment per stacking step, and the throughput of the bonder is the gating variable for the entire fab output. A 12-Hi HBM4 stack requires 11 bonding operations; a 16-Hi HBM5 stack requires 15. ASMPT's TC bonder family directly competes with Hanmi's TC Bonder 4 for the HBM4 socket and has been winning incremental share at SK Hynix through 2025-26 as Hynix has deliberately diversified away from Hanmi's sole-supply position.
The technical specifics of ASMPT's TC bonder positioning matter for the share-loss/share-gain narrative. ASMPT's HBM-class TC bonder platform supports both MR-MUF (the mass-reflow molded underfill process that SK Hynix favours) and TC-NCF (the non-conductive film process that Samsung and Micron historically used) in a single tool — a capability parity with Hanmi's TC Bonder 4. The qualification process at SK Hynix for HBM4 took roughly fourteen months of engineering engagement, which is itself a moat against further competitive entry: any new vendor seeking to compete at SK Hynix's HBM4 line in 2027-28 is structurally late. ASMPT has converted that qualification into recurring tool orders and is now positioned for the HBM4E generation that ramps in late-2026 to mid-2027.
The strategic significance for ASMPT is that the AI-memory ramp is the principal demand driver for its back-end equipment business through this cycle. Hanmi's share loss is ASMPT's share gain — Trendforce reported in December 2025 that ASMPT now operates roughly half of the active HBM4 TCBs at SK Hynix, up from effectively zero in 2024. That share migration is what is showing in the bookings number.
The financial print
Q1 2026 (reported April 29) printed revenue +32% YoY, bookings +72% YoY, and the headline structural metric: SK Hynix HBM4 TC bonder order confirmed as taken from Hanmi. Q2 guidance is $540-600 million in revenue, +37% YoY, above consensus, with the bookings velocity suggesting Q3 and Q4 should sequentially accelerate. Operating margin trajectory is improving from mid-teens toward the 20s as the higher-ASP TC bonder mix flows through.
The stock closed HKD 176.20 on May 17. Q2 2026 reports July 28, 2026.
The bookings-to-revenue lag is the key metric to track. ASMPT's bookings +72% YoY at Q1 will convert to revenue over the subsequent two to four quarters depending on tool type and customer delivery schedule. That suggests Q3 and Q4 2026 revenue can run materially above current consensus, with full-year revenue plausibly $2.2-2.4 billion against consensus closer to $2.1 billion. The book-to-bill ratio at Q1 was running well above 1.0, which by itself is the predictive signal for sequential revenue acceleration. The Q2 print on July 28 should confirm whether the bookings acceleration is sustaining or whether Q1 was a one-time pull-forward.
Customer mix
Customer mix has flipped favourably. SK Hynix is now the principal HBM customer (the share-take from Hanmi), Micron is a smaller but ramping account, and Samsung remains a longer-tail option pending HBM4 qualification. The non-HBM business — SMT lines for general electronics assembly, wire bonders for legacy packaging — provides diversification and represents probably 40-50% of the total revenue line. Per Trendforce's December 2025 reporting, ASMPT was operating roughly half of the active HBM4 TC bonders at SK Hynix's M14/M16 fabs by year-end, up from effectively zero in early 2024 — the share-shift velocity has been faster than any equipment vendor share-shift in recent memory.
The customer book in HBM is structurally improving. Once an HBM line is built out with ASMPT TCBs, the recurring revenue (consumables, service, upgrades) plus the next-generation tool replacement cycle creates an annuity. The customer mix is moving toward higher-quality (HBM, advanced packaging) and away from cyclical commodity assembly.
The geographic distribution of the customer base is part of the strategic story. ASMPT is Hong Kong-listed with Singapore operational headquarters and engineering footprint in multiple Asian semiconductor hubs. This geopolitically-distributed profile is itself an advantage in a world where US-China semiconductor decoupling continues — ASMPT's tools are not subject to US export controls in the same way Hanmi's Korean-origin tools or BESI's European-origin tools might be, and the Hong Kong listing gives the equity a regulatory profile that is distinct from Korean or European peers. As capital allocators increasingly think about supply-chain resilience, that geographic distribution becomes a small but meaningful positive feature of the ASMPT exposure.
Competitive context
The TC bonder competitive set is narrow: Hanmi (#1, ~71% share globally per TechInsights 2025), ASMPT (#2, share rising), Hanwha Semitech (#3, Korean local share-taker, small base), and the longer tail of Japanese assembly-equipment vendors. ASMPT's strategic advantage is the Hong Kong corporate structure (geopolitically more neutral than Korean), the broader back-end portfolio (which is a cost-of-ownership argument for fabs that want one-vendor consolidation), and the hybrid-bonding roadmap that runs through the same corporate parent as the TC bonder business. The Samsung-captive SEMES TCB programme is an internal subsidiary rather than an open-market competitor, so the addressable competitive market is effectively narrowed to the three commercial vendors plus the Japanese long tail.
Against BESI (the hybrid-bonding incumbent at TSMC for logic SoIC), ASMPT is the credible second in the hybrid-bonding race with a different customer footprint. Against Hanmi, ASMPT is the share-taker at the marginal HBM4 socket. The competitive context is structurally favourable in the medium term — ASMPT is moving up the share curve, not defending share.
The recent SK Hynix HBM4 wafer-bonder patent context is informative. SK Hynix filed and was granted patents in early 2026 covering specific wafer-bonding process steps that ASMPT's tool family is positioned to deliver. The implication is that SK Hynix has effectively co-developed elements of the ASMPT HBM4 production flow — a deeper engineering relationship than the typical customer-vendor arrangement and a moat that competitors cannot replicate quickly. The Hanmi share-loss at SK Hynix is therefore not principally a pricing or commercial-relationship issue; it is a process-integration depth issue that ASMPT has earned through the qualification cycle.
Terminal risk
The terminal risk is the same as for Hanmi but in reverse: hybrid bonding eventually displaces TC bonding for HBM5/HBM6 generations in the 2027-29 window. ASMPT has a hybrid-bonding roadmap through the same corporate parent that owns the TCB business, so the displacement risk is partial rather than total — but the value capture per HBM line in hybrid bonding may not match the value capture in TCB. The mitigation is that ASMPT's hybrid-bonding qualification track at SK Hynix and Samsung is concurrent with BESI's, which positions the company for the transition rather than against it.
The secondary terminal risk is HBM cycle peak. ASMPT's bookings are running +72% YoY off a constrained base; the mathematics of base effects mean FY28-29 will deliver structurally lower growth even if absolute volumes hold. The terminal risk is positioning rather than fundamental.
A third structural risk is the non-HBM portion of ASMPT's business — the wire bonder, die-attach and SMT lines that account for roughly 40-50% of revenue. These lines are cyclical and exposed to broader semiconductor capex cycles, automotive electronics demand, and consumer-electronics assembly volumes. If global semiconductor capex outside HBM softens — which is a plausible 2027 scenario given the inventory positions at the end-customer level — the non-HBM revenue line could compress 10-15% in a single year, partially offsetting the HBM tool-order strength. The consolidated revenue trajectory depends on the HBM acceleration outpacing the non-HBM cyclical compression.
Bull case
The three-to-five-year bull case sees ASMPT capturing 40-50% of new HBM4 TC bonder sockets globally through 2027, FY27 revenue running HKD 22-25 billion versus FY26 closer to HKD 17 billion, operating margin expanding to the mid-20s, and EPS at HKD 6-8. A 25-30x multiple — TC bonder duopolist with hybrid-bonding optionality — gets the stock to HKD 200-250 over twelve months.
The stretch bull case adds hybrid bonding qualifying at SK Hynix for HBM5 in 2027 and ASMPT taking principal supplier position alongside BESI, plus continued share gains at Samsung HBM4 production. That path takes the stock to HKD 280-320 on a 24-month view.
A particular dynamic to watch is the Samsung HBM4 qualification path. Samsung has historically been TCB-supplied principally through its in-house SEMES subsidiary (~13% global TCB share) plus a marginal slice from Hanmi. If Samsung's HBM4 generation qualification at Nvidia clears in late 2026 and Samsung's HBM4 capacity expansion runs aggressively in 2027, Samsung will need external TCB capacity beyond what SEMES can supply — and ASMPT is the most likely beneficiary because of its broader process technology and its already-qualified position at the SK Hynix tier. That Samsung optionality is worth 10-15% of incremental revenue in the bull-case path; consensus has not yet modelled it.
Gap / bear case
What the market may be missing is the structural durability of the share gain. Once an HBM line is qualified on ASMPT TCBs, the cost of switching back to Hanmi (re-qualification, line downtime, process validation) is prohibitive — these design wins are sticky 5-7 year revenue annuities, not contestable quarterly orders. Consensus FY27-28 revenue probably understates this durability.
The bear case the market may be right about is the cycle peak — ASMPT's bookings inflection is concurrent with Hanmi's, BESI's, Lumentum's and the entire AI capex complex, and any blink in 2027 hits the whole sector simultaneously. Sizing has to respect that correlated downside.
A specific gap in consensus modelling is the operating-margin trajectory. ASMPT's historical margins have run in the 10-15% range through cycle averages; the current mid-teens margin reflects mix-shift toward HBM-class higher-ASP equipment. If the HBM mix continues to grow as a share of revenue through 2027-28, the consolidated operating margin can structurally lift to the 20-25% range — a 500-1000 bp improvement that would materially re-rate the multiple. Most sell-side models do not yet bake in that mix-shift trajectory at full magnitude.
Optionality
Two options. First, hybrid bonding — ASMPT has the roadmap and the customer engagement, and an HBM5 qualification win would be a step-change re-rating. Second, Singapore corporate-restructuring optionality — the ASMPT parent structure has historically created sum-of-parts value that occasionally surfaces through dividend programs, spin-offs or strategic transactions; none of which is priced into consensus. A third option is the AI-3D-packaging logic-stacking adjacency — as TSMC, Samsung and Intel scale 3D logic packaging for AI accelerators, the TC bonder qualification at the logic-packaging tier becomes a separate revenue line that ASMPT could enter alongside its HBM-focused business; the cross-over from HBM to logic-packaging is a 24-36 month strategic opportunity that is not currently in consensus.
The trade
Entry: HKD 165-185 — current zone acceptable; aggressive add on any 15% pullback to HKD 150. Size: 3% portfolio target (already owned 10K sh per v3 watchlist). Stop: HKD 125 (200-day MA / prior breakout retest). Catalyst date: Q2 print July 28, 2026. Trim/exit: trim 25% at HKD 230, 50% at HKD 270; full exit on HBM cycle peak in monthly Trendforce data. Conviction: 8/10.