Nomura Report: Advanced Packaging Equipment Expanding from CoWoS to Hybrid Bonding — Who Benefits Most?
Nashnova编辑部
Nomura Orient International's machinery team concluded that advanced packaging equipment demand is driven by both 2.5D/3D capacity expansion and per-line value uplift from process upgrades, with beneficiaries falling into short-, mid-, and long-term tiers.
TSMC runs three packaging routes in parallel — why don't equipment makers have to pick sides?
TSMC's 3DFabric spans three routes: CoWoS (2.5D heterogeneous integration), SoIC (3D chip stacking), and COUPE (opto-electronic vertical integration).
This means → the three routes stack rather than substitute — SoIC-processed chiplets still enter the CoWoS flow, and COUPE brings silicon-photonics engines into the same architecture.
In plain terms = equipment makers don't need to bet on a winner. Each new route adds a layer of demand; the routes don't cannibalize each other.
CoWoS packages keep getting bigger — which equipment does that pull?
CoWoS-L — embedding a local silicon interconnect bridge (LSI) inside an RDL interposer — has entered mass production at 3.5× reticle area; long-term scenarios approach 9.5× reticle.
Larger packages directly lift demand for RDL lithography, electroplating, PVD, etching, TCB bonding, and large-area inspection equipment.
This means → every generation of CoWoS-L area expansion pushes both volume and precision requirements for these six equipment categories — the most certain near-term equipment upside.
Bonding equipment: will TCB and hybrid bonding replace or coexist?
TCB thermo-compression bonding is the current earnings driver: 12-layer HBM still uses TCB; CoWoS die-attach for HBM and logic chips relies heavily on TCB; some reflow-soldering steps are also migrating to TCB.
Hybrid bonding is the mid-to-long-term upside lever: as HBM moves to 16 and 20 layers, shrinking micro-bump pitch runs into shorting, warpage, and height constraints. Hybrid bonding eliminates solder and raises interconnect density.
In plain terms = the next several years will more likely see TCB growing alongside capacity expansion, while hybrid bonding enters first in high-end logic, SoIC, and select memory — not a sudden replacement of the old technology.
Lithography equipment: does bigger packaging raise the value of direct-write?
TSV, bumping, and RDL steps in advanced packaging all require patterning. The market currently runs on mask-based lithography; Canon holds a strong position in back-end packaging exposure.
Direct-write lithography needs no fixed mask and can correct for the actual warpage of each panel or wafer. This means → the larger the package and the more pronounced the warpage, the higher the relative value of direct-write.
In China, Mloptic (芯碁微装) equipment is already used in mass production of CoWoS-L-class products at multiple packaging houses; on the mask-based side, Xinyee (新上微装) holds meaningful share in Chinese wafer-level packaging exposure tools.
Panel-level packaging and glass substrates: three layers, very different commercialization stages
As package size keeps growing, the area utilization and size ceiling of round wafer carriers become constraints — driving the industry to explore panel-level packaging.
Glass-related routes break into three layers: ① glass temporary carrier (for panel-level flatness) → ② glass-core packaging substrate (requiring TGV through-holes, metallization, etc.) → ③ glass interposer partially replacing silicon interposers.
In plain terms = the three layers differ sharply in technical difficulty and commercial readiness. Saying "glass substrates are in mass production" lumps them together — only the first layer is close to commercialization today.
China is accelerating expansion — why do equipment orders run ahead of packaging revenue?
SMIC (中芯国际) is building a silicon-interposer and advanced-packaging R&D platform; SiCarrier (盛合晶微), JCET (长电科技), TFME (通富微电), and Silead (甬矽电子) are expanding 2.5D/3D capacity through IPO proceeds, convertible bonds, and project announcements — multiple plans in the tens-of-billions of RMB range.
SiCarrier's IPO case shows roughly RMB 75 billion of a RMB 84 billion project budget allocated to equipment — nearly 90%. This reflects how heavily advanced packaging lines depend on front-end-grade tools, precision die-attach, and inspection systems.
This means → once expansion plans enter real construction, equipment orders appear well ahead of packaging revenue — the core logic behind equipment profit leverage potentially exceeding packaging output growth. Whether hybrid bonding can move from single-tool qualification to large-scale volume is the key mid-term variable to watch.
Content is for reference only, not financial advice.