AI Chip Demand Drives New Growth Cycle for Taiwan's OSAT Companies
N.R. Finch
IDC raised its 2026 global OSAT revenue forecast to 16% year-on-year growth, as advanced-packaging overflow orders shift structurally to independent test-and-assembly houses — ASE and KYEC both posted record May revenues, confirming the cycle has moved from expectation to delivery.
What does "Foundry 2.0" actually mean?
IDC's "Foundry 2.0" label describes a structural shift: independent OSAT firms now absorb overflow advanced-packaging orders from leading foundries, becoming a value-creating link in the chip supply chain rather than a back-end afterthought.
This means → the role of test-and-assembly houses is upgrading. IDC estimates this model will account for roughly 25% of industry output, second only to foundry services at 59%.
In plain terms = when TSMC and peers cannot fit all advanced-packaging work onto their own lines, orders flow systematically — not sporadically — to firms like ASE.
How strong were ASE's May numbers?
ASE Technology (2311.TW) reported May consolidated revenue of NT$63.03 billion (≈US$2 billion), up 28.57% year-on-year — the company's fourth-highest single month ever and its strongest in 43 months.
The real driver is ATM — assembly, test, and materials: May ATM revenue hit NT$42.16 billion, up 37.9% year-on-year, accounting for over 65% of group revenue. This means → growth is powered specifically by advanced packaging and testing, not other business lines.
ASE has already raised its 2026 revenue target from US$3.2 billion to US$3.5 billion — more than double 2025 actual revenue of US$1.6 billion. Management's explanation: advanced-packaging and test demand exceeded expectations.
Why did KYEC also hit a record?
KYEC posted May consolidated revenue of NT$3.78 billion, up 36.61% year-on-year — a single-month all-time high. Year-to-date revenue rose 37.71%.
This reflects a demand source that is easy to overlook: yield challenges in AI chips are rising. In plain terms = AI chips are growing more complex, requiring more pre-shipment testing to screen out defects — burn-in testing (powering and heating chips to simulate long-term use and weed out flaws) and system-level testing orders are surging as a result.
KYEC expects AI-related revenue to climb from 2024 levels to two-thirds of total revenue by 2027 — that target is itself the key checkpoint for whether this demand wave solidifies into a structural business shift or remains a short-term order peak.
Beyond the leaders — how much can the rest capture?
IDC's forecast of 16% annual growth is an industry-wide figure, not exclusive to ASE and KYEC. Other Taiwan OSAT firms — including Chipbond, SIGURD, King Yuan, ChipMOS, and Powertech — are similarly expected to benefit from this upcycle.
This means → AI chip demand for advanced packaging and high-end testing is pulling the entire Taiwan OSAT ecosystem into a growth trajectory, not just the top two names.
The real open question is durability. KYEC's 2027 AI-revenue target (two-thirds) is the most concrete mid-term anchor available — if that number is met, it signals demand has settled into structural business weight rather than a single order peak.
Content is for reference only, not financial advice.