Industry Analysis
ASE's NT$244B quarter is not a cyclical spike—it is a structural repricing. AI silicon has hit a lithography cost wall, and the industry's answer is more packaging, less etching. Chiplet, HBM stacking, and 2.5D/3D integration are shifting value capture from the front-end to the back-end, turning OSAT into the new bottleneck. That is the structural explanation behind the 44.84% YoY jump.
Upstream, ASE's FCBGA and HBM-attach buildout is stretching ABF substrate (Ibiden, Unimicron), underfill (Henkel, Namics), and test-handler (Advantest) order books beyond 12 months. Per-unit test time on AI accelerators runs 3-to-5x that of conventional logic, structurally lifting test-segment margins.
On compliance, ASE's mainland China fabs face BIS export-control exposure on advanced packaging tools. US hyperscalers are pushing OSAT partners to qualify non-China capacity; Vietnam and Malaysia lines still need 12-to-18 months to reach yield parity, creating a capacity-gap window.
The real competitor is not Amkor—it is TSMC. CoWoS internalizes packaging into the foundry P&L, compressing independent OSAT pricing power. ASE's moat: multi-die heterogeneous integration and mature-node packaging—work TSMC will not touch.
12-to-24-month outlook: HBM4 and chiplet complexity will outpace capacity additions. OSAT ASPs should rise 15-to-25%. Packaging premium becomes a standalone BOM line item in AI accelerators, mirroring HBM today. NT$244B is the floor, not the ceiling.
This page displays AI-generated summaries and metadata for research purposes. Original content belongs to the respective publishers.