Bullish on MLCC Industry for Years, 6% Supply Gap Expected by 2028

Overview

J.P. Morgan released an in-depth research report on the Asian MLCC industry, launching its first global long-term MLCC supply-demand model and maintaining a multi-year bullish view on the industry. Core predictions: Driven by dual demands from AI servers and automotive electronics, the MLCC industry will gradually shift from a ~10% supply surplus in 2025 to a ~6% supply shortage in 2028. This level of tightness will slightly exceed the 2017-18 cycle (then driven by cloud capex and early-generation EV demand). The report expects the industry average selling price (ASP) to rise by ~50% during 2026-2028, comparable to the increase in the 2016-18 cycle. The total addressable market (TAM) for the industry is projected to grow at a compound annual growth rate (CAGR) of 24% from 2025-28, exceeding $30 billion by 2028. Based on valuation attractiveness, the report explicitly prefers Japanese (Murata > Taiyo Yuden > TDK) and Taiwanese (Yageo) companies over Korea's Samsung Electro-Mechanics.

Core views

The core arguments of the report are built on three key judgments. Demand Side: AI servers have become the most important application area for the MLCC industry. The report notes that capacity allocation requests from customers have surged for both AI servers and general-purpose servers. Meanwhile, automotive MLCC demand remains relatively resilient. Although end-demand for consumer electronics remains challenging, customers (including distribution channels) have begun to notice tightening supplies. The report predicts that the MLCC industry TAM will achieve a CAGR of 24% from 2025-28 (compared to 16% in 2016-19), with the contribution from ASP increases being more significant. Supply Side: A key point underestimated by the market is the increased trade loss ratio and yield challenges brought about by the rising proportion of server/automotive MLCCs. Nominal capacity in the MLCC industry has grown at an annual average of ~10% over the past decade, but effective capacity after production losses and yield adjustments appears quite limited relative to growing procurement demand. Automotive and server MLCCs use larger package sizes compared to IT applications, and higher system power voltage requirements necessitate the use of more advanced specifications. In particular, the emergence of ultra-high capacitance (e.g., 1005-47uf) new specification MLCCs in AI server applications has led to decreased assembly yields among leading suppliers. This capacity erosion is accelerating—the negative impact on nominal capacity growth will be double-digit percentages over the next three years, whereas it was only low-to-mid single-digit percentages over the past five years. Investment Conclusion: MLCC suppliers have already risen 268% year-to-date in 2026 (vs. MXAP index +16%), significantly outperforming the tech hardware sub-sector. The industry's total market capitalization reaches $354 billion, with consensus EPS-based P/E ratios of 61x/40x/30x for 2026-28E. The report believes that due to the early stage of the industry's transition to the AI ecosystem, consensus earnings revisions often lag, and investors will focus more on growth momentum. Stock price momentum is expected to remain strong in the second half of 2026.

Analysis framework

J.P. Morgan's research methodology revolves around its newly constructed global MLCC supply-demand model. This model integrates historical data updates and a three-year forward-looking outlook, with the following core analytical framework: First, the demand side employs a bottom-up teardown approach for end devices, modeling MLCC demand by application area (mobile, PC/server, consumer electronics, automotive, industrial, and others) before aggregating to the industry level. Specifically, the report separates AI servers from traditional servers to capture incremental demand from GPU and ASIC camp customers. Second, the supply side not only tracks nameplate capacity but crucially introduces the concept of effective capacity—actual output after deducting trade losses from the increasing proportion of large-size MLCCs and adjusting for yields based on nominal capacity. This method is the core innovation distinguishing this report from market consensus. Finally, the report anchors its judgment on the current cycle by comparing the supply-demand tightness and ASP increases of historical cycles (particularly 2016-18 and 2017-18). On the valuation front, it uses multi-dimensional indicators such as PE, PB, and ROE for cross-regional and cross-cycle comparative analysis.

Methodology notes

• Industry/Industrial Analysis FrameworkSupply-demand framework

  Effective Capacity Analysis

  When analyzing manufacturing industries like MLCC, one cannot look solely at nameplate capacity but must consider the impact of product mix shifts (e.g., increasing proportion of large sizes) and yield fluctuations on actual output. In this report, the assembly yield of ultra-high-capacitance MLCCs for AI servers is only 1/3 to 1/6 of ordinary MLCCs, causing effective capacity to be significantly lower than nameplate capacity, which is a key variable in judging the supply-demand gap.

• Industry/Industrial Analysis FrameworkVolume-price decomposition

  Volume-Price Decomposition and ASP Driver Analysis

  The report decomposes industry revenue growth into 'volume' and 'price' dimensions. Of the 24% CAGR for the MLCC industry from 2025-28, the contribution from ASP increases is greater than in previous cycles, reflecting structural price hikes driven by the rising proportion of high-end products (AI servers, automotive) rather than pure volume expansion.

• Valuation MethodPE/PEG valuation

  Cross-Market Valuation Comparison and Growth Premium

  When comparing Japanese, Taiwanese, and Korean MLCC companies, the report looks not only at static PE but also focuses on PEG (Price/Earnings-to-Growth ratio) and valuation repair space. Japanese companies (e.g., Murata, Taiyo Yuden) enjoy valuation re-rating potential when growth certainty improves, due to technological leadership and AI server share advantages; whereas Korea's Samsung Electro-Mechanics sees reduced relative attractiveness due to already high valuations.

• Cycle and Prosperity FrameworkProsperity Inflection Point Analysis

  Superimposed Judgment of Inventory Cycle and Capacity Cycle

  As a typical cyclical industry, MLCC prosperity judgment requires distinguishing between the inventory cycle (short-term, 1-2 quarters) and the capacity cycle (medium-to-long term, 2-3 years). Currently, we are in the upswing phase of the capacity cycle driven by AI demand, while inventories were depleted in 2023-24, creating a 'double-click' effect when superimposed.

Asset mapping & comparison

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

• Murata Manufacturing (Murata, 6981 JP)

  Japanese MLCC leader, technological leader in AI server MLCC, top recommendation in the report

  Strengths

  Technological leadership, highest AI server share, covers leading GPU/ASIC manufacturers, solid automotive MLCC share

  Comparison

  Top pick among the three Japanese firms, highest technical barriers, strongest certainty of benefiting from AI servers

  Risks

  AI server demand falls short of expectations, capacity expansion progress slower than expected

• Taiyo Yuden (Taiyo Yuden, 6976 JP)

  Second largest Japanese MLCC maker, strong technical strength, valuation relatively attractive compared to Murata

  Strengths

  Technical strength second only to Murata, active layout in AI servers, fast growth in automotive MLCC

  Weaknesses

  Smaller scale than Murata, slightly slower AI server customer certification progress

  Comparison

  Second recommendation among the three Japanese firms, larger room for valuation repair

  Risks

  AI server share increase falls short of expectations

• TDK (6762 JP)

  Third largest Japanese MLCC maker, diversified business diversifies risk

  Strengths

  Diversified business (sensors, thin films, etc.), high share in automotive MLCC

  Weaknesses

  Relatively low proportion of MLCC business, slower layout in AI servers

  Comparison

  Third recommendation among the three Japanese firms, relatively smaller growth elasticity

  Risks

  Non-MLCC businesses drag down overall performance

• Yageo (Yageo, 2327 TW)

  Taiwanese MLCC leader, enhancing mid-to-high-end share through M&A integration

  Strengths

  M&A integration effects emerging, product portfolio upgrading, increasing proportion in industrial and automotive sectors

  Weaknesses

  Technical strength still lags behind Japanese leaders, limited AI server share

  Comparison

  Only recommendation from Taiwan, valuation attractiveness higher than Korean peers

  Risks

  M&A integration falls short of expectations, technology upgrade progress slower than expected

• Samsung Electro-Mechanics (SEMCO, 009150 KS)

  Korean MLCC leader, large scale but relatively lagging in AI server layout

  Strengths

  Large overall scale, high share in IT MLCC, strong capital strength

  Weaknesses

  Insufficient technology reserves for high-end AI server MLCC, valuation already high

  Comparison

  Report is less bullish, believing Japanese and Taiwanese companies offer better valuation attractiveness

  Risks

  Loss of AI server market share, pressure from valuation correction

Key data

• 2025-28E Industry TAM CAGR24%Compared to 16% CAGR in 2016-19
• 2028E Industry OPM38%Compared to 36% in 2018
• 2026-28E ASP Increase~50%Comparable to the increase in the 2016-18 cycle
• 2025 Supply Surplus10%Shifts to 6% shortage in 2028E
• AI Server MLCC Assembly Yield1/3-1/6 of Ordinary MLCCKey capacity constraint factor
• Total Industry Market Cap$354 Billion268% YTD gain in 2026
• FY26-28E Consensus PE61x/40x/30xOverall industry valuation level
• Negative Impact of Trade Loss on Capacity GrowthDD% (Double-digit percentage)Next three years; low-to-mid single-digit % in past five years

Impact & implications

The report argues that the MLCC industry is currently in a structural upcycle driven by AI server demand. Compared to the 2017-18 cycle, the supply-demand tightness in the current cycle will be slightly greater, but the ASP increase will be comparable. As a new SKU, AI server MLCCs have technical barriers (side-gap construction method, ultra-high capacitance) and yield challenges, meaning only a few suppliers can achieve mass production. This will limit short-term supply elasticity, supporting prices and margins. Regarding the competitive landscape, Japanese companies (especially Murata) are poised to capture the largest share in the AI server MLCC segment, leveraging technological leadership and close relationships with top GPU/ASIC customers. Taiwan's Yageo holds a position in the mid-to-high-end market through M&A integration and product upgrades. In contrast, although Korea's Samsung Electro-Mechanics has a larger overall scale, its technology reserves and customer certification progress in the high-end AI server market are relatively lagging, and its current valuation already reflects considerable optimistic expectations, thus limiting upside potential. From an investment perspective, the report advises investors to watch for the lagging effect of consensus earnings revisions—as the industry's transition to the AI ecosystem is still in its early stages, market pricing for future growth may still be insufficient. Stock price momentum is expected to continue in the second half of 2026, but close tracking of supplier capacity release progress and secondary market price trends is needed as leading indicators of contract price changes.

Risks

• AI server demand falls short of expectations, leading to a narrowing supply-demand gap and weak ASP increases
• AI server-grade MLCC yields improve faster than expected, leading to unexpected release of effective capacity
• New entrants or existing manufacturers expand capacity too quickly, leading to a resurgence of supply surplus
• Slowing automotive electronics demand weakens the second-largest growth driver
• Macroeconomic downturn leads to further contraction in consumer electronics demand, dragging down overall MLCC demand
• Geopolitical risks affect the stability of the Asian supply chain

What to watch

• Actual progress of supplier capacity expansion and yield improvements
• MLCC certification and order allocation status for AI server customers (GPU/ASIC camps)
• Secondary market price trends as leading indicators of contract price changes
• Pace of recovery in IT and consumer electronics end-demand
• Automotive MLCC demand, particularly changes in EV penetration rates