AI-driven copper-clad laminate, PCB and copper-foil supply chain Report Interpretation

Overview

This deep-dive examines how AI servers and networking equipment are reshaping the upstream PCB materials chain. Morgan Stanley expects a durable but highly selective upcycle for qualified suppliers of high-end CCL, HVLP copper foil and advanced glass cloth, driven by rising material content, lengthy qualification cycles and constrained effective capacity.

Core views

Morgan Stanley’s central argument is that AI is changing the PCB supply chain through specifications, not simply through a rise in unit shipments. Faster data rates, greater power density, higher layer counts and more complex boards require lower-loss CCL, smoother HVLP copper foil and higher-specification glass cloth. The report forecasts global CCL TAM to rise from US$19B in 2025 to US$47B in 2030, a 20% CAGR, versus market expectations of US$35-40B. AI and data-center applications are expected to contribute US$26B of the US$28B total market increase, or 90% of incremental growth, and grow from 23% to 63% of CCL demand. The global PCB TAM is forecast to rise from US$58B to US$135B over 2025-30, an 18% CAGR; AI/data-center PCB demand is projected to climb from US$13B to US$86B and reach 64% of total industry revenue by 2030. The report emphasizes that material content rather than system volume is the main source of growth. It estimates that content accounts for 84% of AI-related CCL growth, as each platform generation requires more complex and higher-layer-count boards as well as higher-grade materials. For Nvidia systems, estimated CCL content per rack rises from US$6,695 for GB300 to US$18,750 for VR200 and US$35,601 for VR300. The same effect appears in other architectures: the report maps copper foil, CCL and PCB content across Nvidia, AMD, TPU and Trainium platforms. Advanced AI PCBs are estimated to require about 102 days, or 14.6 weeks, to produce versus 46 days, or 6.6 weeks, for a conventional multilayer board, reflecting longer glass-cloth production, CCL processing and especially PCB fabrication. Supply is constrained because nameplate capacity is not equivalent to qualified high-grade supply. Moving from lower-grade CCL to M8, M9 or higher requires simultaneous upgrades in resin, copper foil, glass cloth, process control and yield. The report notes that M8 CCL ASP is more than double that of M7, while the CCL ASP index rises from 1.0x for M2 to around 17x for M10. Customer qualification for M7-and-above materials commonly takes 12-18 months and must prove performance in a finished board across production lots. This creates design-in persistence for incumbents, although each platform transition can reopen competition. Morgan Stanley therefore views the cycle as a stock-selection opportunity: commodity CCL may remain oversupplied while qualified high-end materials gain pricing power and margins through 2027-28. Copper foil is presented as the binding upstream constraint. Morgan Stanley forecasts the HVLP4 market to grow from less than US$50M in 2025 to US$2.8B by 2030, a 123% CAGR. Its bottom-up model forecasts monthly HVLP4 demand of 1,028 tonnes in 2026, 3,270 tonnes in 2027 and 4,335 tonnes in 2028, compared with modeled average monthly global capacity of 1,080, 2,495 and 3,600 tonnes. This implies undersupply beginning in late 2026, widening to 31% in 2027 and remaining at 20% in 2028. The report attributes the gap to specialized surface-treatment equipment, slower throughput, lower yields and qualification requirements. Mitsui Kinzoku is estimated to hold 41% of 2026 global HVLP4 capacity and 33% in 2028; Co-Tech is expected to rise from 13% in 2026 to 21% in 2028. Morgan Stanley initiates Overweight on EMC, TUC, Co-Tech and KB Laminates. EMC is identified as the high-speed CCL leader, with capacity forecast to grow from 5.85 million sheets per month at end-2025 to 11.25 million at end-2028. The report forecasts a 76% revenue CAGR and 117% net-profit CAGR over 2025-28, with gross margin rising from 29.8% to 41.4%; its NT$8,400 target implies 20x CY28 P/E. TUC is expected to benefit from high-end networking and AI ASIC exposure, with M7-and-above mix rising from about 36% of sales in 2025 to 46% in 2028, capacity reaching 5.3 million sheets per month and gross margin expanding from 22.7% to 36.0%; the NT$2,700 target implies 17x CY28 P/E. Co-Tech is positioned as a leading advanced copper-foil supplier and a key second source to Mitsui. It redirected capacity away from conventional HTE and standard RTF in 3Q26 toward advanced RTF and HVLP products. Morgan Stanley forecasts HVLP4 capacity of about 200 tonnes per month at end-2026, 700 tonnes at end-2027 and 800 tonnes at end-2028, supporting gross-margin expansion from 21.8% in 2025 to 47.2% in 2028 and an EPS CAGR of 96% over 2025-28. Its NT$730 target implies 23x CY28 P/E. KB Laminates is differentiated by vertical integration across copper foil, glass yarn and cloth, resin and paper-based inputs. The report forecasts its first 350,000 sheets per month of M6+ capacity in 1H27 and another 350,000 in 1H28, alongside a 21,000-tonne high-frequency and low-loss copper-foil plant from mid-2027. It forecasts FY25-28 revenue and net-profit CAGRs of 42% and 91%, respectively, and assigns an HK$75 target, implying 14x CY28 P/E; however, it characterizes KB as more cyclical than Taiwanese high-end peers because commodity FR-4 remains central to its mix. For Shengyi Technology, Morgan Stanley retains Equal-weight despite improving fundamentals, citing valuation. It raises 2026 and 2027 EPS estimates by 52% and 79%, respectively, raises the target price from Rmb74 to Rmb170, and introduces 2028 EPS of Rmb6.98. The report expects CCL and PCB demand, pricing and potential PTFE adoption to support share gains, but notes raw-material cost headwinds and considers valuation fair. Technology choices remain important. PTFE could offer the best electrical performance for future AI networking architectures, but its processing, drilling, lamination and qualification challenges leave hybrid PTFE and M9 alternatives credible. The report says PTFE adoption for Nvidia Rubin Ultra remains unconfirmed. Shengyi could benefit if PTFE-based CCL is adopted, while EMC may compete through a hybrid PTFE solution but could face a smaller addressable market if it sources PTFE externally; TUC has relatively limited direct Nvidia exposure and is viewed as less exposed. More broadly, co-packaged optics could eventually reduce the need for certain premium electrical interconnects, though Morgan Stanley considers this a multi-year rather than near-term risk.

Analysis framework

Morgan Stanley combines bottom-up estimates of CCL, copper foil and PCB content per AI rack or switch with platform shipment assumptions, supply-chain checks and supplier capacity analysis. It then assesses suppliers using six factors: pricing power, AI mix, capital efficiency, market concentration, supply tightness and China substitution risk. Company targets use residual-income models, supported by earnings, mix, margin and capacity forecasts.

Methodology notes

• Industry AnalysisSupply-demand framework

  Bottom-up HVLP4 and high-end CCL supply-demand modeling

  The report compares estimated demand from AI racks and switches with qualified effective capacity, rather than relying on nameplate capacity, to assess scarcity and pricing power.

• Industry AnalysisVolume-price decomposition

  Content-led market growth analysis

  The report separates growth from system volumes and rising material content per system, concluding that specification-driven content growth accounts for most AI-related CCL expansion.

• Competition & strategyEconomic Moat and Competitive Advantage

  Qualification and yield advantage

  Long customer qualification cycles, accumulated process knowledge and reliable yields are treated as advantages that can sustain supplier positioning, while not guaranteeing future wins.

• Valuation methodsRIM (Residual Income Model)

  Residual-income valuation

  For covered companies, the report values beginning equity plus the present value of earnings above the cost of equity, using explicit cost-of-equity, medium-term growth and terminal-growth assumptions.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Elite Material Co. (EMC, 2383.TW)

  Initiated at Overweight as a leading high-speed, low-loss CCL supplier benefiting from AI-server demand and capacity expansion.

  Strengths

  Leading position in high-speed CCL, broad AI platform exposure, capacity expansion and forecast gross-margin expansion from 29.8% in 2025 to 41.4% in 2028.

  Weaknesses

  Addressable market could shrink if PTFE materials are sourced externally in a hybrid PTFE solution.

  Comparison

  The report assigns EMC a higher CY28 P/E than TUC due to its leading technology and production position.

  Risks

  Weaker AI demand, slower material upgrades, faster CPO adoption and Chinese suppliers gaining advanced CCL share.

• Taiwan Union Technology Corporation (TUC, 6274.TWO)

  Initiated at Overweight as a high-speed CCL supplier with networking, ASIC-server and thick-copper CCL exposure.

  Strengths

  M7+ mix expansion, growing AWS Trainium exposure, networking franchise, capacity growth and margin expansion from 22.7% in 2025 to 36.0% in 2028.

  Weaknesses

  Behind EMC in high-end CCL positioning and dependent on new capacity ramps and customer wins.

  Comparison

  The report values TUC at 17x CY28 P/E versus EMC at 20x, reflecting EMC’s stronger technology and capacity leadership.

  Risks

  Slower data-center build-outs, delayed AI/networking platforms, no new CSP customer win and potential AWS Trainium share loss.

• Co-Tech Development Corp. (8358.TWO)

  Initiated at Overweight as an advanced copper-foil supplier and a key second source in HVLP4.

  Strengths

  Qualification moat, withdrawal from low-margin conventional foil, HVLP4 capacity ramp and expected share increase to 21% by 2028.

  Weaknesses

  End-platform exposure is indirect through laminators; production is concentrated at a single Taiwan site.

  Comparison

  Mitsui Kinzoku remains the technology and capacity leader, while Co-Tech is described as the fastest follower.

  Risks

  Weaker server and switch demand, delayed HVLP4 migration, faster competitor capacity additions and rising Chinese competition.

• Kingboard Laminates Holdings Ltd. (1888.HK)

  Initiated at Overweight as a vertically integrated CCL producer benefiting from glass-cloth pricing and a move into higher-grade CCL.

  Strengths

  Control of copper foil, glass yarn and cloth, resin and other inputs improves supply security, cost control and production planning.

  Weaknesses

  Less than 5% of capacity is allocated to M6+ products and commodity FR-4 remains its core market.

  Comparison

  The report considers KB more cyclical and earlier in high-end qualifications than Taiwanese peers EMC and TUC.

  Risks

  Failure to gain high-end CCL share, falling standard-grade prices and competition from leading incumbents or new Chinese entrants.

• Shengyi Technology Co Ltd. (600183.SH)

  Covered at Equal-weight; earnings estimates and target price are raised on stronger CCL and PCB demand.

  Strengths

  Expected AI-server CCL share gains and potential benefit from PTFE-based next-generation CCL.

  Weaknesses

  The report views valuation as fair despite stronger earnings forecasts.

  Comparison

  The report cites Shengyi’s CY28 valuation at 21x versus EMC’s 12.8x CY28 P/E at the time of analysis.

  Risks

  Raw-material cost increases, rising competition, delayed 5G deployment and US-China trade tensions.

Key data

• Global CCL TAMUS$19B in 2025 to US$47B in 203020% CAGR; above market expectations of US$35-40B.
• AI/data-center CCL TAMUS$4B in 2025 to US$30B in 203047% CAGR; rises from 23% to 63% of total CCL demand.
• Global PCB TAMUS$58B in 2025 to US$135B in 203018% CAGR.
• HVLP4 marketLess than US$50M in 2025 to US$2.8B in 2030123% CAGR from a small base.
• HVLP4 supply deficit31% in 2027 and 20% in 2028Morgan Stanley forecasts undersupply beginning in late 2026.
• Advanced AI PCB production time102 days versus 46 days for a conventional PCBAbout 2.2x longer.
• EMC target priceNT$8,40020x CY28 P/E; Overweight.
• TUC target priceNT$2,70017x CY28 P/E; Overweight.
• Co-Tech target priceNT$73023x CY28 P/E; Overweight.
• KB Laminates target priceHK$7514x CY28 P/E; Overweight.
• Shengyi Technology target priceRmb170Raised from Rmb74; Equal-weight.

Impact & implications

The report expects qualified suppliers with advanced specifications, dependable yields and secure access to constrained inputs to capture mix, price and margin gains through 2027-28. It does not view this as a broad commodity-materials recovery: standard-grade CCL can remain price-sensitive or oversupplied while high-end CCL and HVLP foil remain tight.

Risks

• A slowdown in hyperscaler capital expenditure, AI workloads or data-center construction could reduce high-end CCL and copper-foil demand.
• Lower yields, slower ramps or execution failures could raise costs and delay expected margin recovery.
• Suppliers may fail to qualify for next-generation platforms, losing share in higher-margin applications.
• Chinese CCL and copper-foil suppliers could expand qualified capacity and pressure pricing, share and margins.
• Shortages of advanced glass cloth, specialty resin or HVLP copper foil may constrain shipments despite strong demand.
• Faster co-packaged-optics adoption could reduce demand for some premium electrical PCB interconnects and material content.
• Faster-than-expected capacity additions or yield improvements could reduce the modeled HVLP4 supply deficit.
• The market could ultimately overshoot if aggressive investment and inventory building lead to future oversupply and price pressure.

What to watch

• Cloud and hyperscaler capital spending, AI rack and switch shipment trends, customer inventory days, lead times, cancellations and PCB production.
• The pace of M7/M8/M9 material migration, customer qualification outcomes and supplier production yields.
• HVLP4 capacity ramps, specialized equipment delivery times and changes in the modeled 2027-28 supply deficit.
• EMC, TUC, Co-Tech and KB Laminates capacity expansion, product-mix progression, pricing and margin realization.
• Whether PTFE, hybrid PTFE or M9 material solutions are adopted for future Nvidia architectures.
• CPO adoption, Chinese supplier qualification progress, and availability of advanced glass cloth, specialty resin and HVLP copper foil.