Editor’s Note: The supply bottlenecks in AI infrastructure are continuing to spread from GPUs, memory, data centers, and power systems down to more fundamental hardware components. Goldman Sachs and Morgan Stanley have recently turned their attention to MLCCs—multilayer ceramic capacitors, long regarded as ordinary passive components.
In AI servers, MLCCs are critical components that stabilize current and filter noise, ensuring the high-speed operation of chips. As Nvidia’s new rack architecture increases MLCC usage per rack, their value is rising rapidly. Goldman Sachs estimates that the MLCC market for AI servers will grow more than fourfold between 2025 and 2030, while industry capacity is expanding at an annual rate of just over 10%, creating a supply-demand imbalance that has become the key driver of this market cycle.
More importantly, the pricing cycle has begun. Japanese leaders such as Murata and TAIYO YUDEN have initiated price increases, and Japan’s export data is now confirming strong demand. For capital markets, the MLCC thesis is straightforward: demand comes from AI servers and premium automobiles, supply expansion is constrained, and rising prices significantly amplify profit elasticity.
From chips to capacitors, pricing power in the AI supply chain is shifting toward more specialized and less visible segments. Whether MLCCs will become “the next memory chip” depends on whether demand for AI servers can continue to materialize; but one thing is certain: this once-overlooked basic component has now reached the starting point of a new cycle of rising volume and prices.
The following is the original text:
Supply bottlenecks in the AI arms race are sequentially unlocking opportunities across various hardware sectors. Following data centers, energy infrastructure, and memory chips as focal points of capital attention, Wall Street giants Goldman Sachs and Morgan Stanley have both pointed to a long-underestimated foundational component in their latest reports: multilayer ceramic capacitors (MLCCs). Both institutions anticipate that MLCCs will become the next key battleground for simultaneous volume and price growth, and this AI-driven growth cycle may be the largest in history.
Goldman Sachs analyst Daiki Takayama noted in a report that the AI server MLCC market is expected to surge from approximately ¥215 billion (about $1.4 billion) in fiscal year 2025 to around ¥920 billion (about $5.8 billion) in fiscal year 2030, more than quadrupling in size with a compound annual growth rate of 34%. Goldman Sachs explicitly stated that the current AI-driven MLCC cycle “will be the largest and longest in history, and we believe we are still in the early stages.”
MLCC: The 'Invisible Heart' Powering AI Servers
A multilayer ceramic capacitor (MLCC) can be understood as an extremely compact, ultra-fast charge-discharge unit. Unlike conventional batteries, which store large amounts of energy and release it slowly, MLCCs store very little energy but can complete charging and discharging in extremely short time intervals—down to milliseconds or even less. Their primary function is to smooth out power fluctuations and filter noise: absorbing instantaneous voltage spikes or rapidly supplying current during voltage dips to provide stable power to sensitive chips and block electrical interference that could disrupt digital signals.
The operational characteristics of AI servers make MLCCs indispensable. When AI models perform large-scale computations, the power demand of the processor can surge instantaneously within microseconds and then rapidly drop back to near zero after computation ends. The power supply system itself cannot respond quickly enough to such drastic fluctuations. MLCCs are typically mounted directly near AI chips, releasing energy instantly during power peaks to prevent server crashes. Since AI chips such as Nvidia GPUs must simultaneously process billions of tasks, a top-tier AI server rack may require up to 600,000 MLCCs working together to maintain system stability.
Goldman Sachs analyst Nelson Armbrust further noted that MLCCs have become the third most expensive component in AI server bill of materials (BOM), after GPUs and memory. The overall MLCC market is currently around $15 billion, with the server-related segment at approximately $1.3 billion and growing at an 80% compound annual growth rate. In contrast, demand growth in other application areas such as automotive and smartphones has significantly slowed. Daiki Takayama expects the cost share of MLCCs in AI server BOM to rise gradually from the current level of about 0.5% to approximately 1%.
Structural supply-demand imbalance: Annual production capacity growth of only 10% struggles to absorb a fourfold demand surge.
The key factor driving market attention is the severe structural imbalance between supply and demand in the MLCC industry. Goldman Sachs analyst Allen Chang explicitly noted that the industry’s annual production capacity growth rate is only slightly above 10%. Additionally, due to heavy reliance on in-house production of equipment and materials, expansion progress is constrained by internal engineering resources and difficult to significantly accelerate. However, the demand surge from AI servers is on an entirely different scale. Goldman Sachs estimates that MLCC demand driven by AI servers will increase by approximately 4.3 times between fiscal year 2025 and fiscal year 2030.
Even more concerning to the market is the continued strong demand for high-voltage, high-capacity MLCCs driven by automotive electrification, with MLCC usage per vehicle still rising. The two major demand pillars—AI servers and electric vehicles—are jointly consuming the already limited new production capacity. As a result, even amid declining consumer electronics demand, relevant customers are actively seeking long-term supply agreements to guard against future shortages.
Signs of market strain are emerging at multiple levels: lead times for high-end MLCCs (high-capacity, high-voltage specifications) have exceeded 20 weeks; spot and distribution channel prices for low-capacity and consumer-grade MLCCs have risen 20% to 40% due to hoarding and duplicate orders; and key raw materials such as nickel and silver remain at elevated prices, putting pressure on costs across various products.
The price increase cycle has officially begun: Japan's two major players lead the way with price hikes, confirmed by official data.
Price signals are rapidly strengthening. The price hikes by Japan’s two industry leaders, Murata Manufacturing and Taiyo Yuden, mark the official start of the MLCC price cycle. Murata has increased prices for MLCCs used in AI servers and high-end automotive applications by 15% to 35%, effective April 1 this year. Taiyo Yuden has also notified customers of price adjustments across multiple product lines starting in May, covering MLCCs, inductors, RF devices, FBAR/SAW devices, and aluminum electrolytic capacitors, citing sustained increases in the cost of various raw materials, including precious metals.
Japan’s Ministry of Finance’s trade statistics released on May 28 corroborate this price increase trend at a macro level. The data shows that in April, the average export price of MLCCs rose 3% month-over-month and 16% year-over-year; export volumes increased by 10% year-over-year; and export value surged 28% year-over-year. Goldman Sachs believes this data confirms the signal conveyed in recent earnings reports from Japanese MLCC manufacturers: all companies confirmed that order momentum remains strong.
From the timeline of the entire AI supply chain, Goldman Sachs’ analysis framework shows that MLCC price increases have significantly lagged behind those of core AI components such as DRAM, NAND storage, ABF substrates, and copper-clad laminates (CCL). Therefore, Goldman Sachs concludes that among all AI components and materials, MLCCs have the longest potential for price increases and the strongest durability. Goldman Sachs has raised its 2026 year-over-year MLCC price change forecast from approximately 0% to a range of 0% to +5%, emphasizing that actual future increases could far exceed this level.
Remarkable profit elasticity: A 5% price increase could boost operating profit by up to 37%
For investors, the profit elasticity resulting from the supply-demand mismatch in MLCCs should not be underestimated. Daiki Takayama estimates that a mere 5% price increase could theoretically boost Murata’s operating profit for fiscal year 2027 by approximately 13% and TAIYO YUDEN’s operating profit by up to 37%.
Goldman Sachs expects Murata's fiscal year 2027 sales to reach JPY 1.05 trillion (approximately USD 6.6 billion), a 13% year-over-year increase; Taiyo Yuden's sales are projected to reach JPY 286 billion (approximately USD 1.8 billion), also up 13% year-over-year. Goldman Sachs maintains its 'Buy' ratings on Murata, Taiyo Yuden, and TDK. Its constructed Asian MLCC thematic stock portfolio has recently begun to strengthen, but still shows significant catch-up potential compared to other popular AI themes.
Morgan Stanley dissects Nvidia’s new rack: growing importance of peripheral components, MLCC usage surges 182%
Another major catalyst comes from Nvidia’s next-generation Vera Rubin AI rack. After disassembling Nvidia’s latest VR200 rack, Morgan Stanley found that the importance of peripheral components is rapidly increasing in the latest BOM.
The value of MLCCs in a single rack has increased from approximately $1,530 in the previous GB300 era to about $4,320, representing a surge of 182%. Although the absolute value of MLCCs remains lower than that of GPUs, memory, and PCBs, their growth rate is exceptionally prominent among peripheral components.
Morgan Stanley’s channel research further shows that MLCC usage on both compute and switch boards has increased significantly, with a more pronounced rise on compute boards. Additionally, the newly introduced BlueField and ConnectX modules will further boost the total MLCC usage per rack. This helps explain the strong current demand for MLCCs in high-end AI servers and has prompted multiple ODMs to actively stock up in preparation for the mass production and delivery of Rubin racks in the second half of 2026.
Morgan Stanley's teardown of the Nvidia Vera Rubin rack reveals the following changes in the value of key components:
Market intelligence indicates that sequential supply bottlenecks in the infrastructure arms race of the AI supercycle have spawned wave after wave of market winners. Goldman Sachs’ latest assessment describes MLCCs as “the new memory chip”—a passive component segment poised at the outset of a period of rising volume and prices.
Driven by exponential demand from AI servers and Nvidia Rubin racks, lead times for high-end MLCCs have exceeded 20 weeks; Japan’s industry leaders have initiated price increases, and official export data remains robust—all signals point to one conclusion: this AI-driven MLCC supercycle has only just begun.