Goldman Sachs sees AI MLCC cycle extending to 2030

Goldman Sachs expects the AI-driven demand for advanced MLCCs to persist until around 2030, longer than the market anticipates, after meeting with Murata Manufacturing's president.

The AI-driven cycle for advanced multilayer ceramic capacitors may extend to around 2030, Goldman Sachs said after meeting with Murata Manufacturing President Norio Nakajima, as power constraints and edge-device growth stretch infrastructure buildout timelines beyond initial estimates.

"AI infrastructure expansion could continue through 2030, with edge devices like autonomous vehicles and humanoid robots adding a second wave of demand," Nakajima said at the May 27 analyst meeting, according to Goldman Sachs.

Goldman maintained its buy rating on Murata with a ¥5,400 12-month target price, placing the stock on its Conviction List. The bank more than tripled its price target for Micron Technology to $1,625 from $535 in a separate note Tuesday, signaling broad conviction in the AI hardware cycle. Murata plans ¥80 billion in additional capital expenditure specifically for high-unit-price AI applications, much of which can be executed by adding steps to existing production lines rather than building new facilities.

For investors, the extended timeline challenges the prevailing view that AI semiconductor component demand peaks around 2028. If MLCC pricing resets upward with each server platform generation — as Murata's strategy of charging more for new bills of materials implies — suppliers could see multiyear revenue growth beyond current consensus estimates.

Pricing Resets With Each Platform Generation

Murata does not raise prices on existing MLCC products for AI servers. Instead, the company sets higher prices for new components as server platforms upgrade annually, effectively resetting the average selling price upward with each generation. Goldman characterized this as a structural price improvement rather than a cyclical markup, suggesting the pricing power stems from the increasing complexity and density of MLCC requirements in each successive GPU platform.

The AI server MLCC total addressable market breaks down unevenly. Low-voltage, high-capacity advanced products command the largest share, with GPU and ASIC motherboards representing the biggest segment, followed by vertical power delivery systems and network switches. High-voltage, high-capacity primary and secondary power supplies account for a smaller portion.

Silicon Capacitors vs. MLCCs: A Fork in the Road

On the competitive front, Murata acknowledged that silicon capacitors — produced by TSMC, Samsung Electro-Mechanics, and Murata itself — hold advantages for embedded substrate applications due to their thinner profile and better surface flatness, which improves circuit connection reliability. But MLCCs remain superior in capacitance density and customer ease of use.

Murata said it is developing MLCC products with high connection reliability to close the gap, suggesting the company sees both technologies as addressable rather than viewing silicon capacitors as an existential threat. Goldman said the comment shows Murata has the flexibility to compete regardless of which capacitor technology wins in specific applications.

The Broader AI Hardware Picture

The MLCC cycle extension thesis arrives as the broader AI trade regains momentum. Micron Technology's market capitalization topped $1 trillion for the first time Tuesday, with shares up nearly 20%, while Marvell Technology rose more than 10% ahead of its earnings report. The PHLX Semiconductor Sector Index gained 5% on the session.

For Murata, the extended cycle means its ¥80 billion AI-specific capex — deployed efficiently by retrofitting existing lines — could generate returns over a longer horizon than initially modeled. Goldman's conviction rating implies the bank sees the current valuation as not yet reflecting the full duration of the AI MLCC opportunity.

This article is for informational purposes only and does not constitute investment advice.