ASE Technology Holding Co. Ltd.: AI/CPO Outlook Exceeds Aggressive Assumptions; OW

Core Thesis

ASE's 2027 LEAP (leading-edge advanced packaging) revenue could reach US$6bn, driven by broader outsourced CoWoS demand and CPO testing wins, well above prior estimates. The stock remains attractive at 18x 2027E EPS, supporting Overweight with a raised target of NT$558.

What the Market May Be Underpricing

Two upside vectors are underappreciated. First, 2027 LEAP revenue is not capped at the US$5.5bn implied by management’s “increment >US$2bn” hint—our bottom-up work now yields US$6bn, factoring in higher wafer sorting share and expanded AMD Venice CPU volume. Second, ASE has secured insertion 5 in CPO optical testing, a new revenue stream beginning in 2028 that we now incorporate into forecasts. The market also underestimates margin trajectory: pricing power and mix shift already lifted 1Q26 gross margin 294bps YoY to 20.1%, with 2Q26 guidance for another 20-100bps Q/Q improvement.

Evidence Chain

1. First-quarter results and outlook beat consensus. Reported EPS of NT$3.24 (27% Q/Q, 49% YoY) exceeded both our estimate (NT$2.54) and consensus (NT$2.81). Gross margin of 20.1% was 110bps above our model and 50bps above consensus, driven by higher utilization and operating efficiency. Second-quarter guidance calls for revenue +7-9% Q/Q and gross margin +20-100bps, despite rising depreciation—implying continued operating leverage.

2. LEAP revenue trajectory accelerates. Management raised the 2026 LEAP target from US$3.2bn to US$3.5bn, confirming our prior bottom-up estimate. For 2027, the implied increment of >US$2bn (from >US$3.5bn) points to >US$5.5bn. We now model US$6.0bn based on three drivers: (a) nearly 100% oS outsourcing for Nvidia and 70-90% for Google TPU, with dollar content per chip ranging US$150-300; (b) AMD’s Venice CPU adopting full CoWoS-like packaging, split 50:50 between SPIL and ASE eventually, at ~US$320 per chip; (c) rising wafer sorting demand (ASE added capex to meet it). Our breakdown: Nvidia contributes 48% of 2027 LEAP (US$2.9bn), Google TPU 23% (US$1.4bn), AMD 21% (US$1.3bn), others 8%.

3. CPO packaging and testing represent a new growth pillar. For packaging (optical engine + switch IC on same substrate), TSMC is resolving warpage issues in 1H26 before handing production to ASE. For testing, ASE is already involved in insertions 3 and 4 (die-level optical testing) and has now secured insertion 5 (final system-level test). We raise 2028 revenue estimates to reflect this, lifting EPS from NT$35.80 to NT$40.85.

4. Margin expansion is structural, not cyclical. IC ATM gross margin is forecast to rise from 25.9% in 2026E to 28.3% in 2027E, driven by product mix (higher LEAP share) and price increases of 5-20% in ATM services. Total operating margin expands from 12.2% to 15.3% over the same period, as opex declines as a percent of revenue (R&D from 4.4% to 3.7%).

Key Risks

• TSMC capacity loosening could reduce overflow outsourcing to ASE, especially if TSMC expands its own CoWoS clean room. Our oS outsource assumptions (70-100% for major customers) could prove too high if TSMC resolves its constraints faster.
• Consumer demand weakness is a factor for the EMS segment, which we model flat to slightly down; a deeper downturn would pressure overall utilization.
• Competition from Chinese OSATs could erode pricing in mature packaging, though we believe ASE’s advanced packaging moat is wide.
• CPO adoption remains pre-volume; delays in wafer-level warpage resolution could push revenue into 2029.

Valuation & Implication

We raise EPS estimates by 12% for 2026, 16% for 2027, and 14% for 2028, lifting our residual-income based price target from NT$408 to NT$558. The target implies 20x 2027E EPS, slightly below +1SD historical (21x). At current NT$488.50, the stock trades at 18x 2027E EPS—reasonable given 60%+ EPS CAGR through 2028. Bull case (NT$655) assumes 24x 2027E EPS on stronger CPO adoption and sustained margin expansion; bear case (NT$355) assumes 13x on a cyclical downturn. We see 14% upside to the base case and recommend Overweight.

Appendix: LEAP Revenue Estimate Summary