Since September, Murata Manufacturing, the global leader in MLCCs, has taken two seemingly contradictory actions: planning a substantial capacity expansion for high-end MLCCs used in AI servers on one hand, and announcing the production halt of certain consumer and automotive MLCC part numbers on the other.
Industry observers believe this does not reflect strategic wavering at Murata, but rather a reallocation of resources—shifting limited capacity away from traditional markets characterized by low margins and sluggish demand, and concentrating it on high value-added sectors such as AI data centers. Notably, in terms of high value-added strategic deployment, other major MLCC manufacturers are highly aligned with Murata, also actively expanding capacity to meet AI demand.
01. Capacity Expansion and Production Halt: Murata's Capacity Reallocation
Recent Japanese media reports indicate that Murata plans to increase its capacity for high-end MLCCs required for AI servers by approximately 20% from current levels in fiscal year 2028 (ending March 2028). Minamide Masanori, Vice President of Murata, stated publicly that the company is evaluating capacity expansion plans for the next three to five years, including plant construction. He noted that this evaluation "will be bolder than ever before," with the construction of large-scale plants at domestic and overseas factories included as a viable option.
Murata's capital expenditure plan for fiscal year 2026 has reached 250 billion yen, with an additional 80 billion yen specifically allocated for expanding server MLCC capacity. Minamide also projected that the growth in MLCC demand in the AI sector will last at least until 2028, and the medium- to long-term market will extend from AI data centers to edge AI devices.
Running parallel to the capacity expansion is the contraction of its product line. According to media reports, Murata issued a product line optimization notice on September 10: it will halt the production of certain MLCC part numbers in fiscal year 2026, covering standard consumer series and specific automotive specification part numbers, involving series such as GRM, GRJ, GRT, GXM, GXT, GCM, GCJ, GCG, and ZRA. Murata has informed customers to complete their written replies by the end of 2027 and place their final orders by the end of March 2028.
The industry believes that the simultaneous capacity expansion and production halt reflect a relatively clear adjustment in Murata's capacity allocation priorities: traditional part numbers with low margins and sluggish demand are giving way to high value-added AI server products.
02. Resonance in Sync: Top Manufacturers Collectively Pivoting Towards AI Servers
Murata's choice is not an isolated case. Top manufacturers such as Samsung Electro-Mechanics, Kyocera, and Taiyo Yuden are also advancing their respective capacity expansion plans during the same period, with highly consistent directions.
Since May 2026, Samsung Electro-Mechanics has disclosed multiple long-term agreements for AI server MLCCs. The latest news shows that Samsung Electro-Mechanics signed a long-term agreement for AI server MLCCs worth 1.0722 trillion KRW (approximately USD 780 million) with an undisclosed global enterprise, with the supply period covering the entire year of 2027. [Related Reading]
This is the largest single MLCC long-term agreement in the history of Samsung Electro-Mechanics. The long-term agreement model itself also reflects the industry's shift from short-term spot trading to long-term capacity locking.
In terms of capacity, the planned capacity of Samsung Electro-Mechanics' new plant in the Philippines is 1.5 times its existing capacity. It is currently in the stage of approval and land acquisition, and is expected to start releasing partial capacity in early 2028. In addition, the company plans to invest 15 trillion KRW in the Busan area from 2026 to 2040 to build high value-added MLCC production lines and an R&D center.
Kyocera plans to invest a cumulative total of approximately 100 billion yen over the seven fiscal years ending in March 2031, specifically dedicated to expanding MLCC capacity for AI servers. The funds will be primarily directed to the Kokubu site of the Kirishima Plant in Kagoshima, with the goal of doubling the MLCC capacity of "Plant 5-1" by fiscal 2027 compared to fiscal 2026, and further tripling it by fiscal 2028.
Taiyo Yuden plans to invest 270 billion yen from fiscal 2026 to 2030 for MLCC capacity expansion. Meanwhile, the company has already been building a new plant in Malaysia. To support the expansion, Taiyo Yuden is also constructing a new barium titanate R&D and production building at its Haruna Plant in Takasaki City, Gunma Prefecture, Japan. Construction is scheduled to begin in November 2026, with completion and operation targeted for 2028.
The capacity expansion directions of top manufacturers are highly consistent, all targeting the AI server sector. The judgment behind this collective move is that the demand for high-end MLCCs in AI servers is long-term and rigid, and MLCCs are shifting from cyclical products to structurally growing products.
03. Price Hike Cycle Begins: Widening Supply-Demand Gap and Rising Prices
As MLCC capacity concentrates on high-end products, coupled with the sustained strong demand from AI data centers, reactions on the price front have already emerged.
In August this year, the latest survey by global market research agency TrendForce showed that Samsung Electro-Mechanics (SEMCO) took the lead in raising its Q4 2026 quotes for OEM and ODM customers, marking the formal entry of the MLCC industry into an upward trend. TrendForce estimates that Samsung Electro-Mechanics will increase the average prices of consumer-grade X5R products for OEM and ODM customers by 25% to 30%, thereby suppressing order intentions and freeing up capacity to expand high-end production lines. For high-end X6S products applied in AI servers, prices will be negotiated with customers, with average increases ranging from 10% to 20%.
TrendForce further pointed out that, benefiting from the spillover effect of medium-to-high capacity consumer-grade orders, quotes from other manufacturers will maintain an upward trend in the fourth quarter, with an estimated average increase of 10% to 20%, primarily targeting medium-to-high capacity consumer-grade products. Manufacturers' capacity utilization rates are increasing simultaneously, with some suppliers approaching 90%. In addition, due to the sustained strong demand for AI data centers and the widening supply gap for optical communication modules, the demand for MLCCs is also tightening, prompting related manufacturers to actively seek support. In contrast, the outlook for the traditional ICT and automotive markets is conservative. Moreover, as the growth momentum of major economies slows down and end-consumer confidence is under pressure, the traditional market in the second half of the year may face a situation where the "peak season fails to peak."
TrendForce stated that as medium-to-high capacity consumer-grade orders spill over to other manufacturers, capacity utilization rates are elevated, and with SEMCO raising its quotes, the MLCC industry has officially entered a price hike cycle. Whether Japanese manufacturers such as Murata and Taiyo Yuden will follow suit has become a key indicator for observing the next wave of price increases. It is estimated that the fourth quarter will continue the trend of a widening MLCC supply-demand gap and rising prices.
High-end capacity expansion, low-end production halt, and rising prices—this round of adjustment currently experienced by the MLCC industry reflects, to a certain extent, that the demand structure is migrating from consumer electronics to AI infrastructure. How far this trend will go may depend on the sustainability of AI computing power investments and the pace of capacity release on the supply side.
#MLCC #SamsungElectroMechanics #AIServers