PDF 原檔：報告_Daiwa_TPCA2026_20261027_original.pdf
圖片清單（已驗證 2026-10-28）
本報告無抽出圖片（attachment 無對應檔），無需驗圖。
原始內容
3
Taiwan Data Centre Hardware
Rating: Positive
Taiwan PCB: Our key takeaways from 2026 TPCA show
Sheng Cheng
(886) 2 8758 6253
sheng.cheng@daiwacm-cathay.com.tw
Thomas Kim
(886) 2 8758 6249 thomas.kim@daiwacm-cathay.com.tw Summary: We attended the 2026 Taiwan Printed Circuit Association (TPCA) Show that was held on 20-22 October. Below, we share our read-throughs from booth visits and key company takeaways. Overall, we remain positive on Taiwan glass fibre, CCL, PCB, and substrate vendors, supported by upbeat company commentary. On pricing, we expect low-end CCL and E-glass price increases to be larger and more sustained than previously anticipated. High-end CCL vendor commentary suggests plans for more meaningful price increases from 2H26 to maintain their products' relative profitability amid sharp increases in low-end CCL prices. We also expect prices for T-glass, ABF, CCL, and substrates to continue rising. On the CCL technology roadmap, we favour Q-glass combined with hydrocarbon resin (HC) over PTFE-based alternatives as the longer-term direction. Lastly, we provide our view on the CCL market competition landscape change after seeing Nan Ya Plastics targeting the high-end market more aggressively. We also provide some of our key highlights from related companies within the Taiwan PCB supply chain. EMC (2383 TT, TWD4,750, Buy [1]), ITEQ (6213 TT, TWD746, Buy [1]), Kinsus (3189 TT, TWD941, Buy [1]) and NYPCB (8046 TT, TWD1,255, Buy [1]) are our top ideas within our coverage.
We attended the 2026 Taiwan Printed Circuit Association (TPCA) Show last week. Below, we share our read-throughs from booth visits and key company takeaways.
Whether to adopt PTFE emerged as a key point of debate at the show.
- During the TPCA show, Taiwan Union Technology (TUC; 6274 TT, TWD1,280, Buy [1]) unveiled its new M10 (Hydrocarbon base) and M11 (PTFE base) materials with glassfabric (Low-Dk2, Q-glass) and glass-fabric-free versions. Nan Ya Plastic (1303 TT, not rated) also unveiled its M11-grade material (ie, NPG-A99) at the show. Based on our understanding, NPG-A99 will continue to a hydrocarbon resin (HC) base, likely combined with Low-Dk2, to enhance CCL performance for high frequency and highspeed signal transmission requirements. However, Elite Material (EMC; 2383 TT, TWD4,750, Buy [1]) remains reluctant to enter the PTFE-based CCL market, citing limited end application. Also, EMC plans to use advanced Q-glass for its M10 CCL. Lastly, we also heard from Panasonic's booth representative that Panasonic (6752 JP, JPY3,687, Neutral [3]; covered by Ayaka Inomata) is pursuing a similar strategy for M9/M10 development, leveraging Q-glass instead of PTFE at this stage.
- While Nvidia (NVDA US, USD216, Outperform [2]; covered by Louis Miscioscia) appears keen to enhance product performance by adopting PTFE, given its very low dissipation factor (Df), PCB manufacturing yields remain a key concern. Accordingly, some leading vendors are broadening their material options (including PTFE-based CCL) to address potential customer requirements. EMC and Panasonic, however, remain firmly committed to a Q-glass-based technology roadmap. Although TUC highlighted progress on its PTFE-based M11 at a technology conference during the show, we believe it also intends to advance its existing HC resin/Low-Dk2 and HC resin/Q-glass roadmap for its next-generation CCL. In our view, its PTFE-based M11 development is intended to preserve flexibility across potential technology pathways rather than replace its existing approach. Nan Ya Plastics's strategy also appears reasonable: it can produce Low-Dk glass in-house but must source Q-glass and PTFE externally. This sourcing structure may explain its continued use of Low-Dk2 with HCresin instead of PTFE for its next-gen products, including M11.
news
27 October 2026
- Conclusion: which technology roadmap will prevail? We believe it is too early to identify a clear winner, as the supply chain continues to improve manufacturing yields for PTFE-based products. Some vendors remain cautious despite PTFE's long history in the industry, as its thermal shrinkage and relative softness still make adoption in highlayer-count (HLC) PCBs challenging. Given the current stance of leading global vendors EMC, Doosan (000150 KS), and Panasonic, together with EMC's execution track record, we currently favour Q-glass-based solutions without PTFE as the more likely direction. Nevertheless, broadening product offerings to address different customer requirements should not be viewed negatively, provided vendors have sufficient resources to support development across multiple technology pathways.
Low-end CCL and e-glass continue to surprise on the upside.
- During the show, we also spoke with Kingboard Laminates (1888 HK, not rated). The company remained positive on the E-glass outlook, although its booth representative provided no numerical guidance for next year following a c.200% price hike in 2026. Kingboard attributed the pricing strength to supply constraints, driven largely by AI server demand. While we highlighted increased glass yarn supply from Chinese peers in 2Q/3Q26, the company argued that the weaving machine capacity remains the bottleneck. It expects to add only c.500 weaving machines in 2026 and 1.2k in 2027 (vs. current installed base of 3.3k). E-glass fabric price increases therefore appear supported by limited weaving capacity and producers shifting capacity from lower-end glass to higher-end products, including Low-Dk, Low-Dk2, and T-glass. Kingboard also noted a similar trend in low-end copper foil, where shortages have recently intensified.
- The read-through from Nan Ya Plastics is similar. The company indicated that its current fibreglass allocation between internal consumption and external sales is 45:55, with the internal share set to rise gradually as it upgrades its product mix. Its CCL portfolio remains concentrated in FR-4 (M2 grade), but it plans to migrate toward M4/M6, supported by long-term contract wins with two major hyperscalers seeking to secure CCL supply. In 2027, M4/M6 products are expected to account for 60-70% of its mix (4m out of 6.5m in capacity), with M2 and below contributing 20%. The remainder would comprise high-end CCL, including M7/M8/M9, reflecting recent development progress. We therefore expect Nan Ya's external thin-type E-glass supply to decline as its M4/M6 mix increases next year.
Competition from Nan Ya Plastics.
- The current CCL shortage creates favourable conditions for Nan Ya Plastics to expand into the high-end CCL supply chain. Whether the company has entered AMZN's T3/T4 supply chain is also a key point of market debate. Based on discussions with its booth representatives, however, we believe management will prioritise upgrading its M4/M6 product portfolio before pursuing the highest-end CCL market more aggressively. On the positive side, favourable supply-demand dynamics have supported Nan Ya's transition from a low-end supplier toward the mid-to-high-end segment, although high-end CCL should still contribute a relatively small share of revenue in 2027. For incumbent suppliers, competition is clearly intensifying, but we expect the near-term impact on high-end peers to remain limited.
- Our read: We expect the impact on EMC and TUC to remain relatively limited in 2027. Competitive pressure could become more meaningful over 2028-2030, although it is too early to assess its magnitude.
Other key takeaways:
- Nitto Boseki (3110 JP, JPY2,942, Neutral [3]; covered by Noritsugu Hirakawa) identified Taiwan Glass (1802 TT, NR) and Grace Fabric Technology (603256 CH, CNY84.95, Outperform [2]; covered by Dennis Ip) as key competitors. Its plan to triple capacity by 2028 remains largely on track. The company also intends to expand capacity further by 2030, although the details have yet to be confirmed.
- Resonac (4004 JP, JPY12,915, Outperform [2]; covered by Hidemitsu Umebayashi) indicated that customers are requesting ABF substrate core thicknesses of 2,0003,000μm, versus its current maximum of 1,600μm, to accommodate larger package sizes. Potential technology pathways include adopting glass cores or ceramic substrates, or stacking additional core layers to enhance rigidity. Based on discussions with booth representatives, substrate makers may move toward multi-core structures, which could increase T-glass consumption. However, technical challenges in enabling signal transmission between core layers remain a potential barrier that would need to be addressed over the next one to two years.
- Mitsui Kinzoku (5706 JP, JPY2,054, Outperform [2]; covered by Shinichiro Ozaki) indicated that, based on its technology roadmap, M10-grade CCL combined with its HVLP5 copper foil should support 448Gbps SerDes transmission. In the near term, the company plans to shift capacity from lower-end to higher-end HVLP products, with a key piece of production equipment scheduled for installation in 2028/29. On pricing, its booth representative suggested that quarterly increases of more than 10% would be appropriate. Regarding copper foil shortages highlighted by other supply-chain participants, the representative indicated that mid-tier products, such as HVLP1 and HVLP2, could face the most severe supply constraints.
- Unimicron (3037 TT, TWD1,060, Buy [1]) appeared to highlight multi-core structures and the embedding of passive components within core layers as key technology trends in ABF substrates.
- Zhen Ding Technology (ZDT; 4958 TT, not rated) indicated that its substrate business has a plan to list on the Hong Kong Stock Exchange in 1H27 under the name Leading Interconnect Semiconductor Technology (LIST). At the show, the company focused primarily on thick HDI boards and highlighted two next-generation technology trends: 1) sequential lamination (M+N) to improve signal integrity and increase routing density within the same footprint; and 2) hybrid HDI/HLC structures, which add HDI layers to thick HLC boards (M+N), as discussed earlier.
In the interests of timeliness, this document has not been edited.