MLCC (Multilayer Ceramic Capacitor)

A multilayer ceramic capacitor, or MLCC, is a tiny passive component that stores charge and smooths voltage in electronic circuits, with hundreds to thousands on a single board. AI servers use far more high-capacitance MLCCs than traditional servers, because their power-hungry accelerators need dense, clean power delivery.

What is an MLCC?

An MLCC is built from many thin alternating layers of ceramic dielectric and metal electrode, stacked and fired into a single block the size of a grain of sand or smaller. Those layers let a very small part hold a useful amount of capacitance, which is why MLCCs are the most common capacitor in electronics: a smartphone holds roughly a thousand of them, and a server holds far more. Their job is mundane but essential, smoothing and stabilizing the power that feeds every chip. The hardest and most valuable MLCCs are the high-capacitance, high-reliability parts used in cars and, increasingly, in AI servers, where the combination of high current and tight space pushes the limits of the technology.

How is an MLCC used in thematic investing?

MLCCs are a pure picks-and-shovels exposure to AI servers, because content per server is rising sharply. Power-hungry accelerators draw large, fast-changing currents that demand dense decoupling capacitance right at the chip, so an AI server needs many more MLCCs than a traditional one. Murata’s president, Norio Nakajima, has quantified the step-up:

“The number of MLCCs used in AI servers is eight times that of traditional servers.”

— Norio Nakajima, President and CEO, Murata Manufacturing (TrendForce, Feb 6, 2025)

That rising content is why the AI capacitors (MLCC) concept exists. The market is a Japan-and-Korea oligopoly: Murata leads, followed by Samsung Electro-Mechanics, whose AI-server MLCC and substrate businesses helped drive Q1 2026 revenue up 17% to a record 3.21 trillion won (Samsung Electro-Mechanics, Apr 30, 2026), and TDK. The supply of high-capacity MLCC has run behind demand, which supports both pricing and the steady rise in MLCC content per AI server, and it is why the makers are expanding capacity. For investors, MLCCs offer a components-level angle on the AI build-out, treated as a distinct AI-supply-chain exposure rather than a generic component, with the caveat that the business remains cyclical and tied to the smartphone and PC markets.

Related terms & concepts

Frequently asked questions

How many MLCCs does an AI server use?

Far more than a traditional server. Murata's president has said an AI server uses about eight times the MLCCs of a traditional server, because power-hungry accelerators need dense, high-capacitance decoupling right next to the chip.

Who makes MLCCs?

The market is a Japan-and-Korea oligopoly led by Murata, the world's largest maker, followed by Samsung Electro-Mechanics, TDK, Taiyo Yuden, and Yageo. High-capacity MLCCs for AI servers are the most demanding and highest-value part of the market.

Sources & references

  1. Murata Predicts AI-server MLCC Demand to Double (AI servers use 8x the MLCCs) · TrendForce, 2025-02-06
  2. Samsung Electro-Mechanics Announces Q1 2026 Business Performance · Samsung Electro-Mechanics Co., Ltd., 2026-04-30