AI Drives Boom in Asia-Pacific Electronic Hardware Supply Chain; Substrates and High-End CCL Face Tight Supply-Demand

Overview

This report provides an in-depth analysis of the spillover effects of the AI wave on the Asia-Pacific electronic hardware supply chain. J.P. Morgan points out that AI is not just a story about core computing chips; as AI server architectures continue to upgrade, demand for peripheral fundamental electronic components (passive components, PCB/CCL, substrates) and testing equipment is experiencing explosive growth. Due to technical barriers and capacity constraints in high-specification products, certain sub-sectors (such as high-grade copper-clad laminates and large-size substrates) are poised to face severe supply-demand imbalances. Asia-Pacific manufacturers with core capacity are entering a golden era of rising volume and prices.

Core views

Passive Components (MLCC, etc.): The Total Addressable Market (TAM) for AI server MLCCs is expected to expand at a Compound Annual Growth Rate (CAGR) of over 150%. As computing power increases, capacity consumption of high-end MLCCs rises significantly. Furthermore, manufacturers like Yageo benefit not only from MLCCs but also possess unique AI growth logic in product lines such as resistors and tantalum capacitors for AI servers, with pricing expectations trending upward. PCB and Copper Clad Laminates (CCL): AI is driving rapid expansion in the High-Performance Computing (HPC) CCL/PCB market. Demand for M8 and M9 grade high-end materials is surging (e.g., UBB/LPU adapted for Nvidia's Vera Rubin architecture), and by 2027, M8/M9 materials are projected to account for the vast majority of the HPC CCL market. Due to yield rates and capacity limits of high-grade materials, supply may fail to meet explosive demand. Additionally, Low Earth Orbit (LEO) satellites represent a significant incremental market for PCB/CCL. Substrates: The industry is entering an unprecedented period of supply tightness. With the drastic increase in size and layer count of AI chips (from Nvidia Blackwell and Rubin to Feynman generations), the substrate area required per chip has increased exponentially. Meanwhile, shortages of upstream T-glass have further exacerbated supply chain bottlenecks, prompting manufacturers and customers to sign Long-Term Agreements (LTAs) to secure capacity. Testing Equipment: Testing equipment manufacturers represented by Chroma are seeing earnings explode. With soaring GPU power consumption (TDP) and the evolution of Co-Packaged Optics (CPO) technology, demand for System-Level Testing (SLT), optical communication testing, and data center power testing has achieved a compound growth rate of over 100%. Chroma holds an absolute leadership position in AI data center power testing, enjoying extremely high gross margins.

Analysis framework

The institution adopted an analytical framework combining 'bottom-up decomposition of AI hardware BOM (Bill of Materials)' with 'supply-demand gap estimation'. By tracking the AI chip roadmaps of leading cloud providers such as Nvidia, Google, and AWS (including chip size, layer count, and power consumption evolution), they quantitatively calculated the multiplier effects on MLCC usage, CCL grades, substrate area, and testing time. Combined with upstream capacity expansion plans and yield bottlenecks, this leads to the conclusion that supply-demand gaps are inevitable in high-end segments.

Methodology notes

• Industry/Sector Analysis FrameworkSupply Chain Upstream-Midstream-Downstream Transmission

  Supply Chain Specification Pull Effect

  By mapping the supply chain transmission path of 'Glass Cloth/Copper Foil -> CCL -> PCB/Substrate -> AI Server/Chip', the report precisely identifies the specification pull effect of AI computing upgrades on upstream base materials (such as M9 grade CCL and T-glass), helping readers understand how terminal demand translates into volume and price dividends for upstream materials.

• Industry/Sector Analysis FrameworkSupply-demand framework

  High-End Capacity Supply-Demand Gap Estimation

  In the substrate and high-grade CCL sectors, the institution estimated not only the demand surge driven by AI (e.g., doubled substrate consumption due to larger chip areas) but also compared global top-tier manufacturers' capacity expansion plans against yield losses. This led to the core judgment that 'supply will fall below demand,' which is key to predicting industry price hikes and manufacturer earnings elasticity.

• Industry/Sector Analysis FrameworkVolume-price decomposition

  Decomposition of TAM Growth Drivers by Volume and Price

  When estimating the market size for MLCC and CCL, the institution decomposed growth drivers into 'shipment growth' and 'ASP (Average Selling Price) uplift from specification upgrades,' revealing the dual boost to supply chain profitability from the increased proportion of high-end products.

Asset mapping & comparison

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

• Yageo (2327.TW)

  AI servers require not only MLCCs but also significantly increased amounts of high-end resistors and tantalum capacitors. The company possesses a complete product line with upward pricing expectations.

  Strengths

  Comprehensive product portfolio, high market share in tantalum capacitors, and economies of scale.

  Comparison

  Broader exposure to AI benefits across passive components compared to pure-play MLCC manufacturers.

• Elite Material (2383.TW)

  Core supplier of HPC CCL for AI servers, deeply benefiting from the explosion in demand for M8/M9 grade high-end materials.

  Strengths

  Leading share among core customers like Nvidia; high technical barriers in high-grade materials.

  Comparison

  Leads peers in AI server CCL market share.

• Unimicron (3037.TW)

  Global substrate leader, directly benefiting from the surge in substrate area demand caused by larger AI chip sizes and resulting supply tightness.

  Strengths

  Large capacity scale, leading technology in high-end ABF substrates, and premium customer structure.

  Comparison

  Ranks in the global top tier in terms of capacity volume and technological reserves.

• Chroma (2360.TW)

  Increased GPU power consumption and adoption of CPO technology significantly extend testing time and boost demand for testing equipment.

  Strengths

  Absolute leadership in AI data center power testing and SLT, with extremely high gross margins.

  Comparison

  Stronger product matrix and market share in specific high-power testing fields compared to overseas competitors.

Key data

• AI Server MLCC TAM CAGR150%+Expected Compound Annual Growth Rate for 2025-2028
• HPC CCL TAMApprox. USD 9.98 billionProjected market size in 2027, with M8/M9 grade materials accounting for 73%
• Substrate Industry Capacity Utilization>100%Expected to exceed 100% in 2027-2028, entering an overloaded state
• Chroma SLT & Power Testing Business Growth100%+Expected compound growth rate for related AI testing business in 2025-2027

Impact & implications

The report suggests that dividends from AI hardware have fully spilled over from core GPU chips to peripheral fundamental electronic components and testing segments. Asia-Pacific manufacturers possessing high-end material formulations (e.g., M9 grade CCL), mass production capabilities for large-size substrates, and monopoly advantages in specific testing fields (e.g., high-power supply testing, SLT) will gain significant pricing power and earnings elasticity. Supply chain tightness will prompt downstream major clients to increasingly adopt Long-Term Agreements (LTAs) to bind quality capacity, further raising industry entry barriers.