High-value-added MLCCs for AI servers and data centers: AI infrastructure demand is reshaping the MLCC cycle toward higher-value, supply-constrained products

Overview

The report examines how AI-server and data-center buildouts are changing MLCC demand from a conventional electronics cycle into a higher-value, technology-constrained growth opportunity. Morgan Stanley expects small, high-capacitance MLCC demand, pricing and supplier profitability to strengthen from 2026 and accelerate from 2027.

Core views

Morgan Stanley argues that MLCCs are still early in the AI infrastructure cycle. These capacitors stabilize voltage, supply instantaneous current and remove noise near GPUs, CPUs and FPGAs. Faster chips, lower operating voltages, rising power consumption and constrained board space increase the need for smaller, higher-capacity, low-ESL MLCCs. The report expects this technical shift to make AI-related MLCC demand structurally more valuable than a conventional cyclical recovery. The institution forecasts global MLCC shipment value to rise at a 20.3% CAGR from US$14.67bn in 2025 to US$44.45bn in 2031, versus a 5.0% CAGR during 2012-25. Its 2026, 2027 and 2028 market-value forecasts were raised to US$18.51bn, US$24.52bn and US$28.60bn, respectively. AI/data-center MLCC demand is forecast to reach US$23.33bn by 2031, up from the prior US$10.80bn estimate, implying a 57.1% CAGR rather than 38.1% previously. The report links this revision to accelerating MLCC content in next-generation AI racks. It estimates VR200 NVL72 rack demand at roughly 570,000 units, around 80% above GB300 NVL72's approximately 320,000 units, while dollar MLCC content per rack rises 166%. High-capacitance products of 47µF and above could comprise 50-60% of incremental Cloud AI MLCC total addressable market in 2027, representing about US$2bn of additional opportunity. The report estimates their Cloud AI volume could rise from around 4bn units in 2025 to more than 40bn units by 2027. It expects 47µF+ usage to exceed 30% in VR200, compared with less than 20% in GB300. Morgan Stanley sees a supply-demand mismatch because high-end AI MLCCs require advanced dielectric materials, highly uniform slurry, precision multilayer stacking, controlled firing and demanding customer qualification. Products can require hundreds to more than one thousand dielectric layers with near-zero defects. The report argues that these barriers make a rapid repeat of the supply response seen in the 2017-18 shortage less likely. Murata and SEMCO together hold roughly 85% of the high-end AI MLCC market; lead times exceed 20 weeks, inventories are lean after two years of destocking, and meaningful new capacity is largely scheduled for 2027 or later. This should support tighter capacity, higher pricing and a richer mix, while potentially crowding out traditional consumer, automotive and industrial applications. Murata and SEMCO are identified as the strongest beneficiaries. Murata held 40.8% global MLCC share in 2025 and SEMCO 22.5%; the report expects both to gain share as AI/data-center demand expands. MLCCs represented 50% of Murata sales and 46% of SEMCO sales in 2025. Morgan Stanley estimates Murata's AI/data-center MLCC sales will grow roughly 120% year on year in F3/27 and about 100% in F3/28, aided by both volume and ASP gains. It forecasts Murata F3/27 operating profit of ¥490.8bn, above guidance of ¥430bn and FactSet consensus of ¥453.9bn, followed by ¥702.6bn in F3/28. The stock remains Overweight and a Japan electronic-components Top Pick, with a ¥11,000 DCF-based target and 53% expected upside over 12-18 months; the target was reduced from ¥12,500 solely mainly because the DCF risk-free rate rose from 2.6% to 3.0%. For SEMCO, Morgan Stanley forecasts operating profit of W2.34tn in FY26, W5.70tn in FY27 and W8.36tn in FY28, supported by MLCC Components and Package Solution growth. Components operating margin is projected to rise from 21.4% in FY26 to 30.9% in FY27 before easing to 30.1% in FY28. The Overweight target price is W2,620,000, based on 13x 2027E P/B, with MLCC pricing, AI content, ABF substrate opportunities and customer diversification as key drivers. Yageo is expected to benefit both from AI-related demand and from capacity conversion away from mid- and low-end products. The report forecasts CY26 revenue of about NT$183bn, up 38% year on year, gross margin of 39.3%, operating margin of 27.4% and EPS of NT$19.79. For CY27, it forecasts NT$252bn of revenue, up 38%, gross margin of 43.2%, operating margin of 33.4% and EPS of NT$32.11. Morgan Stanley notes that a 2Q26 pricing surge was partly driven by a temporary inventory build, which later normalized and contributed to a roughly 50% share-price retreat from about NT$1,140. Nevertheless, it considers the MLCC upcycle intact and expects high-end capacity conversion to tighten mid- and low-end supply from late 2026 into early 2027. It remains Overweight, though CY26-28 EPS estimates were cut by 2%, 6% and 7%, respectively, and the residual-income-based target was reduced from NT$1,465 to NT$1,050. Taiyo Yuden is upgraded from Underweight to Equal-weight after its share price fell 56.0% from the July 1 peak, but Morgan Stanley sees only 6% upside over 12-18 months and lowers its DCF target from ¥12,500 to ¥10,500. The report expects earnings growth, forecasting F3/27 operating profit of ¥45.1bn and F3/28 operating profit of ¥66.1bn, but believes the company trails Murata and SEMCO in small, high-capacity MLCCs for AI accelerator boards. Its MLCC market share is estimated to have fallen from 12.7% in 2020 to 11.3% in 2025, and to 10.7% in 1Q26 and 9.9% in 2Q26. The report highlights risks to the industry thesis. AI/data-center capital expenditure could slow because of power-infrastructure bottlenecks or financing practices involving cloud or semiconductor providers. Slower AI investment would leave MLCC demand below the base case. It also flags the risk of slower order-intake growth after the sharp rise since July-September 2025, macroeconomic and inventory-cycle exposure across consumer electronics, supply execution, foreign-exchange sensitivity and weaker end demand. For individual suppliers, risks include market-share loss, increased competition and localization, weaker smartphone or consumer demand, inventory corrections, and slower progress in high-value MLCC technology.

Analysis framework

Morgan Stanley combines AI-server component-content estimates, supply-chain checks, MLCC market-value forecasts, supplier market-share and earnings models, pricing and utilization analysis, and company-specific bull/base/bear scenarios. It then applies DCF or residual-income valuation frameworks to the covered suppliers.

Methodology notes

• Industry AnalysisSupply-demand framework

  AI-driven demand growth versus constrained high-end MLCC supply

  The report assesses how rising AI-server content, long qualification cycles, limited supplier concentration and delayed capacity additions may tighten supply and support pricing.

• Industry AnalysisVolume-price decomposition

  MLCC volume, ASP and product-mix analysis

  It separates shipment growth from higher ASPs and explains how the shift toward smaller, high-capacitance MLCCs can raise market value and margins.

• Valuation methodsDCF (Discounted Cash Flow)

  DCF valuation for Murata and Taiyo Yuden

  The report discounts projected cash flows using stated risk-free rates, beta, risk premium, WACC and terminal-growth assumptions to derive target prices.

• Valuation methodsRIM (Residual Income Model)

  Residual income valuation for SEMCO and Yageo

  The framework values forecast returns earned above the cost of equity, together with continuing value, to establish base-case equity value.

Asset mapping & comparison

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

• Murata Manufacturing (6981.T)

  Primary beneficiary of AI/data-center high-value MLCC demand

  Strengths

  Top global MLCC share, technology leadership in miniaturization and high-capacity products, and expected AI/DC MLCC sales growth.

  Weaknesses

  Broad exposure to consumer electronics and inventory cycles.

  Comparison

  Expected to benefit more than Taiyo Yuden and, with SEMCO, dominate high-end AI MLCC supply.

  Risks

  Failure to meet rapidly rising AI/DC demand, macro weakness, inventory cycles and yen appreciation.

• Samsung Electro-Mechanics (009150.KS)

  Primary beneficiary of AI MLCC and ABF substrate demand

  Strengths

  Second-largest MLCC supplier, concentrated high-end positioning, AI-driven mix improvement and Package Solution opportunity.

  Weaknesses

  Exposure to mobile demand and execution in Chinese smartphone customers.

  Comparison

  Together with Murata, holds around 85% of the high-end AI MLCC market.

  Risks

  Market-share decline, greater competition and localization, weak consumer demand and a Samsung flagship-cycle pullback.

• Yageo Corp. (2327.TW)

  Beneficiary of MLCC utilization, pricing and capacity-crowding effects

  Strengths

  Expected operating leverage from utilization and firmer pricing, with potential expansion into AI applications.

  Weaknesses

  Recent price increases partly reflected a one-off inventory build and estimates were reduced.

  Comparison

  Lower direct high-end AI exposure than Murata and SEMCO but positioned to benefit from tighter mid- and low-end supply.

  Risks

  AI-demand reversal, inventory risk, consumer-electronics weakness, macro deterioration and competitive pricing pressure.

• Taiyo Yuden (6976.T)

  Participant in MLCC recovery with more limited AI upside

  Strengths

  Core MLCC and inductor earnings are projected to grow; the share-price decline limits downside in the report's view.

  Weaknesses

  Declining MLCC share and weaker positioning in small, high-capacity AI-server MLCCs relative to Murata and SEMCO.

  Comparison

  Expected to benefit later and less materially from AI/data-center premium MLCC demand than Murata and SEMCO.

  Risks

  Weak global growth, lower high-end consumer production, inventory-cycle sensitivity and volatile MLCC pricing.

Key data

• Global MLCC shipment value CAGR20.3% in 2025-31Forecast to rise from US$14.67bn in 2025 to US$44.45bn in 2031.
• AI/data-center MLCC marketUS$23.33bn by 2031Revised from US$10.80bn; implies a 57.1% CAGR.
• VR200 NVL72 MLCC contentApproximately 570,000 unitsVersus around 320,000 for GB300 NVL72; dollar content per rack is estimated to rise 166%.
• Murata and SEMCO high-end AI MLCC shareAround 85% combinedA concentrated market structure supporting the report's supply-tightness thesis.
• Murata F3/27 operating profit forecast¥490.8bnAbove ¥430bn guidance and ¥453.9bn FactSet consensus.
• SEMCO FY27 operating profit forecastW5.70tnComponents, Package Solution and Optics forecasts imply substantial earnings acceleration.
• Yageo CY27 revenue forecastNT$252bnUp 38% year on year, with projected operating margin of 33.4%.

Impact & implications

Morgan Stanley believes AI infrastructure spending shifts MLCC demand toward technologically advanced, higher-margin products, favoring suppliers with high-end capacity, manufacturing capability and established customer qualifications. It expects Murata and SEMCO to capture the largest direct benefits, Yageo to gain from broader supply tightening and pricing, and Taiyo Yuden to participate in growth but lag the leading suppliers in premium AI applications.

Risks

• AI/data-center capital expenditure could slow because of power-infrastructure bottlenecks or changes in AI financing and cloud-service arrangements.
• A slowdown in order-intake growth after the recent sharp increase could weaken market expectations.
• High-end MLCC suppliers may fail to meet demand or lose market share despite a favorable market backdrop.
• Consumer, automotive and industrial demand weakness, inventory corrections and macroeconomic deterioration could pressure broad MLCC demand and pricing.
• Competition, China localization, foreign-exchange movements and slower technology progress could weaken supplier earnings or valuation outcomes.

What to watch

• AI/data-center capex growth and the pace of GPU and rack-generation transitions, particularly VR200 NVL72 deployment.
• Demand, ASP and capacity utilization for high-capacitance 47µF+ MLCCs.
• High-end MLCC lead times, long-term supply agreements, capacity additions and pricing trends.
• Murata and SEMCO AI-related MLCC sales growth, market-share trends and margin progression.
• Yageo inventory normalization, utilization, direct-customer pricing and evidence of supply tightening from late 2026 into early 2027.
• Taiyo Yuden's market-share trajectory and premium AI/data-center MLCC contribution.