China advanced packaging, CCL and AI-related PCB and copper-foil supply chain: Morgan Stanley favors China advanced-packaging beneficiaries as AI demand expands CCL, PCB and copper-foil markets

Overview

Morgan Stanley's Greater China technology webcast examines how AI infrastructure demand could reshape advanced packaging, CCL, PCB and copper-foil markets. The report expects strong growth in AI-linked materials and packaging, prefers equipment vendors to OSATs within advanced packaging, and differentiates between JCET's expected share gains and SJ Semiconductor's already-demanding valuation.

Core views

Morgan Stanley expects China's advanced-packaging market to reach Rmb100bn in 2029, with chiplet 2.5D packaging expanding at a 40% CAGR. The firm expects annual domestic 2.5D packaging capacity to rise to 436k wafers in 2027. This capacity build supports suppliers of packaging equipment, which Morgan Stanley considers better positioned than OSAT operators to capture the investment cycle. At the same time, the capacity outlook leads the report to expect moderate overcapacity in China's 2.5D packaging market: CoWoS-S capacity may become excessive, whereas CoWoS-L may remain tight. Within OSATs, Morgan Stanley expects JCET to gain share in AI GPU packaging as its computing and automotive mix continues to rise. The report rates JCET Overweight and sets a Rmb90.2 price target. It identifies better-than-expected communications, computing and consumer-electronics demand, earlier-than-expected provision of 2.5D/3D packaging, and faster share gains as upside factors; weaker end demand, delayed 2.5D/3D delivery and slower share gains are the corresponding downside risks. SJ Semiconductor is described as a leading advanced-packaging participant, but Morgan Stanley believes its premium valuation already discounts technology leadership and AI-driven growth. SJ Semiconductor is rated Equal-weight with a Rmb118.4 price target. Its outlook is tied to progress in China's front-end leading-node yields, local and overseas HBM availability, and advances in 3D packaging. The report also argues that AI and data-center hardware materially increases demand for the electronics interconnect stack. AI/data-center-related CCL TAM is projected to grow at a 47% CAGR from 2025 to 2030 and account for 63% of CCL mix by 2030. Global PCB TAM is expected to grow at an 18% CAGR over 2025-30, while AI/data-center-related PCB TAM grows at a 45% CAGR and reaches 64% of PCB mix by 2030. The mechanism is greater material content per AI rack: advanced AI PCBs take 2.2 times as long to produce as normal PCBs, and projected rack-level copper foil, CCL and PCB content rises substantially across NVIDIA, AMD, TPU and Trainium platforms. For copper foil, Morgan Stanley forecasts global R-PCB copper-foil TAM growth of 14% CAGR from 2025 to 2030, with AI/data-center applications reaching 58% of the total market by 2030. It expects the high-performance HVLP4+ segment to be supply constrained, estimating an undersupply gap of approximately 31% in 2027 and approximately 20% in 2028. The report links this shortage to demand from AI accelerators, custom AI chips and high-speed networking applications that require higher-grade foil.

Analysis framework

The report combines market-size and capacity forecasts with supply-demand analysis across advanced packaging, CCL, PCB and copper foil. It then links AI-server platform specifications and rack-level material content to downstream demand, while using company-specific technology, customer, share-gain and valuation considerations to differentiate JCET and SJ Semiconductor.

Methodology notes

• Industry AnalysisSupply-demand framework

  Capacity and supply-demand analysis for China's 2.5D packaging and HVLP4+ copper foil.

  Morgan Stanley compares expected capacity expansion with demand to identify moderate 2.5D packaging overcapacity and projected HVLP4+ foil shortages.

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  AI compute-platform demand is traced through packaging, CCL, PCB and copper-foil requirements.

  The report uses rack-level material-content estimates to explain how AI accelerators and data-center systems increase demand for upstream interconnect materials.

• Valuation methodsRIM (Residual Income Model)

  Residual income model used to derive company price targets.

  The model values long-term earnings after charging for equity capital; the report specifies cost-of-equity, growth and terminal-growth assumptions for its target-price calculations.

Asset mapping & comparison

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

• JCET Group Co Ltd (600584.SS)

  Covered advanced-packaging company expected to gain share in AI GPU packaging.

  Strengths

  Expected 2.5D packaging share gains and a growing computing and automotive mix.

  Comparison

  Morgan Stanley favors JCET's expected share-gain profile relative to the report's more valuation-cautious view on SJ Semiconductor.

  Risks

  Weaker demand, delayed 2.5D/3D packaging capability, or slower-than-expected share gains.

• SJ Semiconductor Corp (688820.SS)

  Covered advanced-packaging company positioned as a technology leader.

  Strengths

  Leading advanced-packaging position and AI-driven growth exposure.

  Weaknesses

  Morgan Stanley believes its premium valuation already discounts its technology leadership and AI-driven growth.

  Comparison

  Unlike JCET, SJ Semiconductor is rated Equal-weight rather than Overweight.

  Risks

  Limited front-end leading-node yield improvement, slower local HBM ramp-up or tighter HBM import restrictions, and limited progress in 3D packaging.

Key data

• China advanced-packaging TAMRmb100bn in 2029Morgan Stanley forecast
• Chiplet 2.5D packaging growth40% CAGRGrowth within China's advanced-packaging market
• China 2.5D packaging annual capacity436k wafers in 2027Expected capacity level
• AI/data-center CCL TAM growth47% CAGR, 2025-30AI/data-center CCL mix projected to reach 63% by 2030
• AI/data-center PCB TAM growth45% CAGR, 2025-30AI/data-center PCB mix projected to reach 64% by 2030
• HVLP4+ copper foil undersupply~31% in 2027e; ~20% in 2028eProjected supply gap
• Advanced AI PCB production time2.2x normal PCBReported manufacturing-time comparison

Impact & implications

Morgan Stanley's analysis indicates that AI infrastructure demand should support higher-value packaging equipment and advanced CCL, PCB and copper-foil demand. However, the report distinguishes between segments with potential oversupply, particularly CoWoS-S, and constrained segments such as CoWoS-L and HVLP4+ copper foil, while emphasizing company-specific execution and valuation differences.

Risks

• China's 2.5D packaging market could face moderate overcapacity, with CoWoS-S potentially becoming excessive.
• JCET could face weaker-than-expected communications, computing and consumer-electronics demand, delayed advanced-packaging delivery, or slower share gains.
• SJ Semiconductor could be affected by limited leading-node yield improvement, constrained HBM supply or limited progress in 3D packaging.

What to watch

• The pace of China's 2.5D packaging capacity expansion and the balance between CoWoS-S and CoWoS-L supply.
• JCET's progress in AI GPU packaging and its computing and automotive revenue mix.
• Local HBM capacity ramp-up, overseas HBM sourcing conditions and progress in 3D packaging.
• Whether AI and data-center demand sustains projected CCL, PCB and HVLP4+ copper-foil growth and supply tightness.