BofA bullish on substrates: server chips add value with each generation, with multilayer cores ahead of glass cores
Following its TPCA research, BofA says substrate specification upgrades and technical barriers support pricing; glass cores offer superior performance, but cracks and reliability still constrain mass production.
Original source：BofA Global ResearchFirst published on Unhidea：Original source date：
Key takeaway
BofA expects substrate suppliers to benefit from server chip specification upgrades. Multilayer cores are the more feasible near-term path, while glass cores face a longer period of preparation for mass production.
Why is BofA more bullish on substrates?
After attending the TPCA exhibition, BofA reiterated that it prefers substrates to copper-clad laminates (CCL) and printed circuit boards (PCB): with each generation of server CPU and GPU upgrades, substrate value rises by about 50%, and suppliers are securing more pronounced price increases than in previous cycles.
BofA and Prismark expect the substrate market to reach 40 billion dollars by 2030; FC-BGA substrates’ share will rise from about 50% in 2025 to about 70%. The substrate industry’s average selling price is expected to grow at a compound annual rate of 9.1% from 2020 to 2030, after declining in each of the previous three five-year cycles.
BofA maintains Buy ratings on substrate manufacturers Unimicron, NYPCB and Kinsus, as well as EMC, which is upgrading to copper-clad laminates for substrates.
Why are multilayer cores feasible earlier than glass cores?
Multilayer cores can help control warpage caused by factors such as mismatched thermal expansion of materials and uneven copper distribution, while improving rigidity and mechanical drilling capability and supporting data signal integrity.
Glass cores have more pronounced long-term advantages: compared with organic cores, their warpage is about one-third as much, their coefficient of thermal expansion is lower, and their rigidity is higher. But cracks may appear only after processes such as thermal cycling, handling or assembly, making them harder to address near the final-product stage; mismatched thermal expansion also creates reliability issues.
BofA’s roadmap predicts that glass cores will become mainstream for large substrates no earlier than 2030; supplier AT&S believes the entire glass-core supply chain is still some way from being ready for mass production. Ceramic cores are constrained by panel size.
Which materials will be used to upgrade copper-clad laminates?
BofA believes a combination of Q-glass fiberglass cloth and hydrocarbon resin is better suited to copper-clad laminate upgrades in the near to medium term, balancing performance and processability. EMC’s M10-grade copper-clad laminate is primarily based on hydrocarbon resin and is scheduled to enter the new product introduction stage in the first quarter of 2027; its M11-grade solution using Q-glass is in early research and development.
PTFE solutions offer strong performance, but have a relatively high coefficient of thermal expansion and weaknesses in copper plating, drilling and lamination. As suppliers place greater emphasis on low thermal expansion, BofA believes PTFE is less likely to become a long-term solution.
What other areas will specification upgrades affect?
The supply chain expects PCB layer counts to rise to 40–50 layers over the next one to two years, with adoption of multilayer cores also increasing, making drilling more critical. BofA observes that high-end equipment and drilling suppliers are paying greater attention to substrate demand; Japanese suppliers are relatively unenthusiastic about expanding capacity, which could prolong the supply shortage for high-end PCBs and substrates.
Substrate manufacturers also have a potential new market: BofA notes that 3.2T optical modules may use the semi-additive process (SAP), potentially expanding substrate manufacturers’ serviceable market over the long term.
