Murata Plans ¥1 Trillion Investment Over Five Years to Expand MLCC Production as AI Data Center Demand Exceeds Expectations
nashnova research
Murata Manufacturing will spend ¥1 trillion over five years to more than double MLCC capacity — driven by AI data-center demand that is outpacing the company's own projections.
¥1 trillion over five years — what exactly is Murata building?
Murata plans to invest ¥1 trillion (~$6.4 billion) to more than double production of MLCCs — multilayer ceramic capacitors, tiny components that regulate current in phones, cars, and servers.
This means → the company sees AI-driven demand not as a short-term spike but as a structural shift worth a strategic-scale bet.
In April the company announced an ¥80 billion two-year add-on, lifting annual capacity growth from ~10% to ~25%. This five-year plan dwarfs that. In plain terms = the last expansion was outdated before the ink dried.
How strong is the demand signal?
CEO Norio Nakajima told Bloomberg this month that component demand is rising faster than projected just one quarter ago — and planned capex is still not enough.
This reflects a notable pivot: three months earlier he had expressed caution about the sustainability of AI data-center investment. He has now shifted from "wait and see" to "chase capacity full speed."
U.S. hyperscalers are pouring hundreds of billions of dollars into data-center networks. The MLCCs they need must be small, durable, and energy-dense — the high-end segment where Murata leads.
What do the latest earnings show?
For the quarter ending September: operating profit hit ¥147.5 billion on revenue of ¥588.5 billion, both well above analyst estimates.
Murata raised its full-year operating-profit guidance to ¥520 billion, citing data-center demand as the core driver of improved sales and margins.
This means → the expansion is not a speculative bet on future orders — it is a response to demand already on the books that current capacity cannot fill.
Why does AI suddenly need so many capacitors?
TrendForce notes that AI workloads are shifting rapidly from model training to inference and AI agents, triggering a new wave of CPU demand.
Each CPU requires power-delivery modules and DDR5 memory, both of which consume high-capacity MLCCs. In plain terms = the more AI moves toward inference, the more CPUs are deployed — and each CPU drags a chain of capacitor demand behind it.
This signals that MLCCs are no longer just "smartphone parts" — they have become a critical bottleneck component in AI infrastructure.
Where is the risk?
Concerns about the financing health of AI data centers have not faded — CoreWeave and other emerging cloud operators have drawn scrutiny for heavy borrowing to buy Nvidia chips.
This means → if downstream customers hit funding problems, the "better-than-expected demand" visible today may not continue in a straight line.
Whether this financing risk turns into an actual demand pullback is the key test for Murata's expansion thesis.
市场有风险，内容仅供研究参考，不构成投资建议。