欣興二季度毛利率大幅超预期并上调2026年资本开支

• Nice GM beat on consensus forecasts. Unimicron 2Q GM was meaningfully above consensus at ~3ppt above the street, which resulted a double digit beat (+20%) on consensus OP. 2Q GM rose ~7ppt vs. 1Q, and we believe the performance was due to 1) previously secured price hikes starting to kick in, and 2) rising substrate rev %, and 3) improving yields in HDI. Meanwhile, the non-op (likely due to strong share price performance in some key Taiwan tech holdings) ballooned the EPS to TWD8.25, far above the TWD3.3-3.5 range of the expectations (but investors likely understand that this may not be sustainable given the recent market weakness). Unimicron is holding a local analyst meeting at 2PM on July 29th.
• Our thoughts on 2Q/2H; mix matters more than hikes. We expect company GM to reach the 30% level in 3Q, while rev to be up in the range of 10-15% q-o-q. We believe ABF rev has benefitted from like-for-like price adjustments since late-1Q, and ABF GM has reached close to 40% by end of 2Q, which is a remarkable improvement from the ~30% in 1Q26. In 2H, we believe mix improvement will steadily take over like-for-like price hikes as the main driving force of the margin expansions. Our non-consensus thesis on pricing (mix matters more than price hikes) has drawn a very high number of investor inquiries since it is published in our preview note (link). We are confident that, with the rapid mix improvement (numerous AI accelerator project launches in 2H), Unimicron will 1) see overall GM reaching close to 40% by YE26 (while ABF GM should reach comparable levels to the 2022 peak, on a recurring basis), and 2) match or even outgrow its 2027 substrate rev vs. peers, which are aggressively seeking like-for-like pricing hikes.
• Raising capex for 2026. The board also approved the updated 2026 capex budget, which was raised from NT$34bn to NT$53.7bn. We believe the substantial increases will be utilized to resolve debottlenecks of critical processes (such as substrate core production) that are seeing rising intensity in the coming generations of leading edge substrate designs. Meanwhile, some of the increased budget could be utilized to speed up the 2027 capacity expansions (likely KF 2, plus the potentially earlier construction start for YM2). We expect the company to share the long term plan (up to 2030) for the ABF substrate capacity expansions. The notable new site would likely be YM3.
Investment Thesis, Valuation and Risks
Unimicron (Overweight; Price Target: NT$1,150.00)
Investment Thesis
We are OW on Unimicron, given the long-term positive ABF substrate demand growth and emerging technological convergence, which we think puts the company in a good position to leverage opportunities such as SLP, Rigid-flex PCBs, etc. For ABF, there is a strong content growth opportunity (~50-100%) from rising layers/areas for HPC/AI/5G chips. The company has also been increasingly prudent regarding execution and capital spending, in our view.
Valuation
Our Dec-26 PT of NT$1,150 is based on an FY27E P/E of 27.5x, which is ~1.5 STD above the average multiple since 2018 (which excludes extremes). We believe this multiple is justified by the growth in key ASIC projects together with the GPU project ramp-up.
Risks to Rating and Price Target
Downside risks to our rating and price target include:
• Threats from alternative packaging technologies, such as silicon interposers and 3D ICs, which could significantly reduce the value-add of substrates.
• ABF capacity expansion by Japanese/Korean peers that more than compensates for demand growth.
• Macro overhang that pressures capital investments and CE/enterprise demand. A muted smartphone outlook could adversely affect demand for SLP/BT.
• Delays in 5G mmWave pull-in demand, which could be negative for the contribution upside from BT.