AI Optical Interconnects Drive PCB Volume and Price Uplift; ZDT Emerges as Top Winner
AI summary card
AI Optical Interconnects Drive PCB Volume and Price Uplift; ZDT Emerges as Top Winner
Morgan Stanley forecasts the AI optical transceiver PCB market to grow at an 83% CAGR driven by technological upgrades. ZDT is poised to capture outsized share leveraging its mSAP process advantage; target price raised to NT$666.
- AI optical transceiver PCB market size projected to expand from US$620mn in CY25 to US$3.77bn in CY28 (83% CAGR)
- 1.6T optical modules to drive PCB ASP from US$10 to US$25, with gross margins potentially expanding to 40-50%
- ZDT viewed as the most attractive name; target price raised to NT$666, implying 20x CY28 P/E
- Unimicron maintained at Overweight with TP raised to NT$1,285; Shennan Circuits maintained at Equal-weight with TP of Rmb400
- mSAP process becomes a core barrier in the 1.6T+ era; ZDT gains first-mover advantage via Apple supply chain experience
- CPO will not disrupt pluggable optical modules in the near term; both are expected to coexist during the 3.2T transition period
Report interpretation
Overview
This report focuses on an undervalued segment in AI infrastructure build-out: optical transceiver PCBs. Morgan Stanley notes that as AI cluster scales expand and optical module speeds evolve from 400G to 1.6T, PCBs are seeing not only shipment growth but also a doubling in unit value due to upgrades in materials, layer counts, and processes. The report highlights Zhen Ding (ZDT) as its top pick, arguing that its expertise in high-end mSAP processes positions it well to gain market share amid industry consolidation. It maintains an Overweight rating on Unimicron but keeps Shennan Circuits at Equal-weight due to valuation concerns.
Core views
AI optical interconnects are becoming an infrastructure bottleneck as critical as compute capacity, creating structural growth opportunities for PCB manufacturers. The report forecasts the global AI optical transceiver PCB market to surge from approximately US$620mn in 2025 to US$3.77bn in 2028, representing an 83% CAGR, significantly outpacing the 60% growth rate of optical module shipments. This excess growth stems from increased 'content': as speeds advance toward 800G and 1.6T, PCB layer counts increase from 10-12 to 14-16, copper-clad laminate (CCL) grades upgrade from M6 to M8, and manufacturing processes shift from HDI to mSAP. Consequently, PCB value per module jumps from ~US$10 to ~US$25, while gross margins are expected to expand from 20-30% to over 40-50%. Regarding the competitive landscape, the report identifies Zhen Ding (ZDT) as the most attractive investment target. While incumbents like Unimicron and Shennan Circuits will also benefit from industry beta, ZDT holds a first-mover advantage in certification barriers for high-end 1.6T+ products, leveraging its mass production experience with mSAP gained from serving Apple iPhone motherboards since 2017. Channel checks indicate ZDT's share in the 800G segment has started at ~7.5% in CY25 and is projected to reach 20% by CY28; in the 1.6T segment, share could rise from 5% to 25%. This share grab driven by technological generational gaps, combined with explosive industry growth, forms the core bullish thesis for ZDT. For the other two covered companies, the report maintains an Overweight rating on Unimicron, raising its target price to NT$1,285, primarily citing its dominance in ABF substrates for AI ASICs and server CPUs, along with expectations of tight supply for high-end capacity post-2027. For Shennan Circuits, while acknowledging its long-term beneficiary status in China's data center and semiconductor localization trends—and having raised earnings forecasts and target price to Rmb400—the report sees limited near-term upside given current valuations of 25x CY28 P/E, thus maintaining an Equal-weight rating.
Analysis framework
The report employs an analytical framework combining typical 'volume-price decomposition' with 'technology iteration mapping'. First, it derives an exponential growth curve for optical module shipments by tracking GPU cluster expansion plans of hyperscalers. Second, it deeply dissects the evolution of technical specifications for PCBs inside optical modules (layer count, materials, processes) to quantify the ASP uplift associated with each generation of speed upgrades, thereby reaching the core conclusion that 'PCB market growth > optical module shipment growth'. In terms of stock selection logic, the firm applies a 'technological barrier-share migration' model. By analyzing the necessity of mSAP processes in the 1.6T era, it uses mass production experience with this process (particularly from high-end consumer electronics applications) as a key metric for screening alpha. On valuation, it adopts a multi-stage Residual Income Model rather than simple PE multiples to more accurately capture the ability of companies to generate excess returns during high-capex cycles, combined with PEG metrics for cross-comparison of value among the three companies.
Methodology notes
Decomposing market size growth into two drivers: shipment volume growth and unit value/content growth
The report notes that the boom in the AI PCB market is driven not just by increased optical module volumes (60% CAGR) but more importantly by doubled ASPs due to rising PCB complexity (ASP rising from $10 to $25). This 'volume and price uplift' logic is a key method for assessing elasticity in niche sub-sectors.
A valuation method based on book value plus the present value of future excess earnings
For asset-heavy PCB/substrate manufacturers in high-capex cycles, simple PE ratios may be misleading. The RI model measures true value creation by deducting equity cost, making it more suitable for evaluating whether such companies can earn returns above their cost of capital during technology upgrade cycles.
Technical entry barriers constituted by specific manufacturing processes (e.g., mSAP)
The report views mSAP process capability as the 'ticket to entry' for the 1.6T era. Manufacturers with mature mass production experience in this process (derived from the Apple supply chain) gain higher customer certification barriers and pricing power, which is the core competitive factor distinguishing Alpha returns from Beta returns.
Adoption speed of new technologies/products in the market exhibits S-shaped acceleration characteristics
The report mentions that the AI optical transceiver market is entering an 'exponential' growth phase, with 1.6T products ramping rapidly in CY26-28. Identifying S-curve inflection points during technological generation switches (400G→800G→1.6T) is a common method for forecasting demand explosions in upstream components.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- Zhen Ding (4958.TW)Top Beneficiary: Poised to capture outsized market share in the 1.6T era leveraging mSAP process advantages and aggressive capacity expansion
- Strengths
- High maturity of mSAP process (derived from Apple supply chain); 40%+ gross margin on AI optical module PCBs; ABF substrate business turns profitable and expands international customers (e.g., Google TPU)
- Weaknesses
- Still significant exposure to iPhone demand; execution risks in new capacity ramp-up
- Comparison
- Fastest share growth in AI optical module PCBs among peers (177% CAGR) and more attractive valuation (15x CY28 P/E vs peer average 25x) compared to Unimicron and Shennan
- Risks
- iPhone sales below expectations; execution failure in Google TPU substrate project; intensified competition from Chinese peers leading to pricing pressure
- Unimicron (3037.TW)Core Beneficiary: Leader in AI ASIC/CPU ABF substrates, also benefiting from volume and price uplift in optical module PCBs
- Strengths
- Dominant position in high-end ABF substrate market; capacity tightness post-2027 supports pricing power; solid share in AI server PCBs
- Weaknesses
- Weak PC demand dragging overall performance; raw material shortages may constrain low-end product shipments
- Comparison
- Higher elasticity in ABF substrate business than ZDT, but slightly slower share growth in the AI optical module PCB niche compared to ZDT
- Risks
- Weakening demand for AI servers and general-purpose servers; yield breakthroughs in alternative technologies (e.g., CoWoP) faster than expected; delays in new capacity commissioning
- Shennan Circuits (002916.SZ)Domestic Substitution Beneficiary: Core play on China data center and semiconductor localization, but valuation fully reflects this
- Strengths
- Leading share in China 5G base station and data center PCBs; rapid growth in IC substrate business; deep beneficiary of domestic semiconductor self-sufficiency trend
- Weaknesses
- Slow recovery in 5G base station demand; elevated current valuation (25x CY28 P/E) limits near-term upside
- Comparison
- Irreplaceable in the domestic Chinese market, but lags Taiwanese peers in globalization and pace of technological iteration
- Risks
- Delays in 5G deployment; escalation of US-China trade friction; intensified competition in data center PCBs
Key data
- AI Optical Transceiver PCB TAM (CY28E)~US$3,773MSignificant increase from US$620M in CY25, reaching 83% CAGR
- 1.6T Optical Module PCB ASPUS$20-30~2.5x increase compared to US$5-15 for 400G
- ZDT CY28E AI PCB Share~20% (800G) / ~25% (1.6T)Significant improvement vs CY25, reflecting mSAP technology dividends
- AI Optical Transceiver Shipments (CY28E)158M unitsIncluding 87M units of 1.6T; 3.2T begins small-batch shipments of 8M units
- ZDT Target PriceNT$666Implies 20x CY28 P/E, 0.5x PEG
Impact & implications
The report argues that AI optical connectivity is reshaping the competitive landscape and value distribution in the PCB industry. For the supply chain, this means manufacturers with high-end HDI/mSAP capabilities and advanced material application experience will achieve profit margins and growth rates superior to the industry average. AI optical transceiver PCBs are poised to become the engine driving a new round of capex and technological iteration for HDI/SLP technology platforms, following smartphones. For investors, this highlights the need to focus on suppliers within the AI hardware chain capable of transitioning from 'pure manufacturing' to 'high-tech barrier components', especially those able to enter top-tier customer supply chains during technological generational switch windows.
Risks
- Demand for general-purpose servers and AI servers weaker than expected
- Large-scale industry capacity expansion plans leading to deteriorating supply-demand dynamics
- Faster-than-expected adoption of CPO (Co-Packaged Optics) technology, impacting demand for pluggable optical modules and their PCBs
What to watch
- Whether AI optical transceiver shipments and 1.6T penetration rates meet quarterly expectations
- Progress of ZDT's certification and mass production with international customers such as Google TPU
- Market price trends for ABF substrates and supply conditions for T-glass materials
- Pace of data center construction in China and implementation of semiconductor localization policies