New Supercycle? Morgan Stanley: MLCC is Turning from a "Commodity" into a "Strategic Resource"

```

The explosive expansion of AI computing power infrastructure is pushing a small ceramic capacitor, previously regarded as an ordinary electronic component, to a strategic core position in the supply chain.

According to Chase Trade Desk, Morgan Stanley points out in its latest research report that Multilayer Ceramic Capacitor (MLCC) is undergoing a structural transformation—from a cyclical commodity into a strategic resource in the AI era, and supply-demand mismatches may persist for several years.

The report notes that a single AI server consumes about 10 to 15 times more MLCCs than a traditional server, while the Nvidia GB300 platform uses as many as 320,000 MLCCs per rack, and the next-generation Vera Rubin platform will see content value increase by about 1.8 times.

Morgan Stanley estimates that AI-driven MLCC demand will surge more than four-fold between 2025 and 2030, while the industry's overall capacity only grows by 10% to 15% per year, with high-end AI-grade capacity effectively booked for several years ahead. Channel surveys show that since Q1 2026, distributors' prices have risen by 200% to 300%, and the spot price for certain specifications has surged by as much as 6 to 10 times.

Morgan Stanley raised the target price for Samsung Electro-Mechanics from 920,000 KRW to 2,560,000 KRW, maintaining an overweight rating. At the time of the report, the MLCC sector had seen a 434% increase year-to-date, far surpassing the Asian tech stock average of 101%, with Samsung Electro-Mechanics up as much as 763% this year.

Unlike 2017 This Time

Morgan Stanley compares the current MLCC cycle with the supercycle of 2017-2018, but emphasizes that today's constraints are more structural.

The 2017 cycle was driven by the smartphone 5G upgrade and EV demand—a cyclical shock from the demand side. The key difference this time is: existing MLCC capacity (including EV-related lines) fundamentally differs from the specifications required by AI servers and cannot be directly converted. AI servers require ultra-low Equivalent Series Resistance (ESR) and Equivalent Series Inductance (ESL) to handle nanosecond-level transient current response, requirements that consumer or automotive-grade products cannot meet.

New MLCC capacity takes about two years from investment decision to mass production, and equipment is highly customized, making efficiency improvements less rapid than in semiconductors. Moreover, manufacturers usually demand sustained price increases to start new capacity investment; most companies set annual capacity growth targets at 10%–15%, deliberately avoiding speculative expansion. High-capacitance, low-ESL AI-grade MLCCs also face stringent qualification barriers, which further restrict competition from lower-end manufacturers.

Morgan Stanley also notes that the rise of Agentic AI is bringing incremental demand beyond GPUs—the CPU-intensive nature of inference computing will further increase MLCC consumption.

AI Servers Reshape Consumption Structure

AI computing platforms' demand for MLCCs is not only reflected in quantity but also in a comprehensive upgrade of specifications.

In terms of use , AI server motherboards are estimated to carry 15,000 to 25,000 MLCCs, about 10 times that of ordinary servers. The Nvidia GB300 platform uses about 320,000 MLCCs per rack, and the next-generation Rubin NVL72 rack will climb to about 570,000 MLCCs, with MLCC content value per rack increasing about 182%.

In terms of specifications, demand is shifting to smaller size, higher capacitance, higher temperature and lower ESL products. MLCCs of 47μF and above are taking an increasing share in AI server bills of materials, exceeding 30% in the Rubin platform while less than 20% in the GB300 system. Morgan Stanley estimates that demand for MLCCs above 47μF in cloud AI will rise from about 4 billion units in 2025 to about 38 billion units in 2030.

In terms of capacity consumption, high-capacitance AI-grade MLCCs require hundreds or even thousands of dielectric layers, so each product consumes far more capacity than its shipment share suggests. Morgan Stanley notes that even if AI-grade MLCCs only account for about 3% of industry shipments and 5% of revenue by 2027, their actual capacity consumption will far exceed these figures, tightening supply across the industry and pushing prices higher.

Price Rises Spread from Spot to Contract Markets

Morgan Stanley's channel survey indicates that multiple substantive signs of supply-demand tension are emerging in the market.

In the spot market, prices in Shenzhen's Huaqiangbei have risen sharply since early May, initially focused on high-capacitance products, then spreading to consumer-grade specifications. Some types (such as 0805 27μF) have risen about six-fold since the start of the year, with some extreme specifications up by 10–20 times. However, Morgan Stanley points out some of these reflect traders' hoarding activity, not end-user demand.

On the contract pricing side, leading suppliers have raised prices by 20%–30% (quarter-on-quarter) for distributors in Q2 2026, with some manufacturers cancelling bulk discounts and stopping low-priced orders from Q1. Distributors expect further price hikes in Q3 and are actively seeking long-term agreements (LTA) to secure supply.

On the inventory side, after two years of destocking, supply chain inventories are at low levels—distributors have about 1.5 months of inventory on average, downstream customers less than one month. Taiwanese MLCC monthly sales data show a 40% year-on-year increase in May, confirming the upward cycle. Lead times for high-end products exceed 20 weeks; pricing logic is shifting from cyclical fluctuation to structural premium.

Japanese and Korean Duopoly Dominates High-End Market

The high-end AI-grade MLCC market is dominated by two companies from Japan and Korea, with a combined market share of about 85%.

Murata holds about 45% market share and is the industry's technical benchmark. Its AI server orders are about twice its capacity, with utilization rates at 90%–95%. Its competitive advantage comes from vertical integration of materials and its leadership in ultra-miniature 008004 packaging. Murata’s president Nakajima says that demand for cutting-edge MLCCs for AI/data centers will significantly exceed expectations for at least the next three years.

Samsung Electro-Mechanics (SEMCO) has about 40% market share, narrowing the gap with Murata by shifting from low-margin consumer electronics to higher-margin AI and automotive fields. Its Tianjin factory is running at full capacity.

Additionally, Chinese manufacturer CCTC (Sanhuan Group) is entering the domestic server supply chain.

Global MLCC Market Value to Accelerate Expansion

Morgan Stanley forecasts global MLCC shipment value will increase from $14.7 billion in 2025 to $24.3 billion in 2028, corresponding to an annual compound growth rate of 18%, significantly higher than the 6.5% historical trend over the past decade. AI and data centers will be the main incremental drivers.

The addressable AI server MLCC market (TAM) is expected to reach about $900 million by 2027, and if AI infrastructure demand continues to surpass expectations, could exceed $1 billion in 2030. Morgan Stanley expects supply tightness to persist into the second half of 2026 and 2027. The ASP increase driven by product portfolio upgrades will be a more enduring industry characteristic than spot price fluctuations.

Morgan Stanley notes that the current MLCC sector's overall price-to-book ratio has risen to 9.6 times, far above the historical average of 1.7 times. The investment logic has shifted from "buying undervalued cyclical stocks" to "accelerating earnings revision and scarcity value." Key catalysts include further contract price increases, sustaining high order-to-shipment ratios, near-full capacity utilization, and final confirmation of MLCC content in the Rubin/VR200 platforms.

~~~~~~~~~~~~~~~~~~~~~~~~

The above highlights are from Chase Trade Desk.

For more detailed interpretation, including real-time commentary, frontline research, and more, please join the Chase Trade Desk Annual Membership

Risk Warning and DisclaimerThe market has risks, so investment requires caution. This article does not constitute personal investment advice and does not take into account any individual user's special investment objectives, financial situation, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article suit their specific circumstances. Investment based on this content is at your own risk. ```