[Edaily Reporter JAEMIN SONG ] Artificial intelligence (AI) is even changing the production priorities on multilayer ceramic capacitor (MLCC) production lines. This follows the decision by Japan’s Murata, the world’s leading manufacturer, to phase out certain general-purpose and automotive products and reallocate production capacity to other products. Just as memory manufacturers prioritized high-value-added products with their limited production capacity following the surge in demand for high-bandwidth memory (HBM), the MLCC market is also seeing a full-scale reorganization of production portfolios centered on AI.

According to industry sources on the 14th, Murata recently decided to phase out certain part numbers from nine MLCC series, including GRM, GRJ, GRT, and GCM. This includes not only general-purpose products for consumer electronics and industrial equipment but also some automotive electronics products. The company has requested that its customers switch to alternative products by the end of 2027, with final orders accepted through March 2028 and shipments continuing through March 2029.

Murata is not discontinuing all nine series. This move is more akin to product portfolio optimization, in which certain part numbers are phased out to expand production capacity for other products. Although Murata has not explicitly stated that it is redirecting this production capacity toward AI servers, the industry views this as a production portfolio adjustment tied to the recent sharp increase in demand for high-specification MLCCs.

This interpretation is driven by the rapidly growing demand for MLCCs used in AI servers. AI servers consume more power and contain more components than standard servers; furthermore, since they operate 24 hours a day, they require MLCCs with high capacitance, high heat resistance, and high reliability. Consequently, in the MLCC market—which had traditionally been centered on home appliances and information technology (IT) devices—AI servers are rapidly establishing themselves as a new, high-value-added market segment.

Differences in profitability are also accelerating the shift in the production portfolio. The securities industry estimates the operating profit margin for MLCCs used in AI servers to be over 30%. This is more than 22 percentage points higher than the 8% margin for general-purpose products; based on the same sales revenue, operating profit is approximately 3.8 times higher.

In fact, orders from leading manufacturers are outpacing shipments. According to market research firm TrendForce, as of the end of June, the book-to-bill (BB) ratio—the ratio of order value to shipment value—for Murata, SamsungElectroMechanics, and Taiyo Yuden stood at 1.30, 1.31, and 1.25, respectively, marking the highest levels since the COVID-19 pandemic. A BB ratio above 1 indicates that new orders exceed shipments. Analysts attribute this to the demand for high-specification MLCCs being driven by the advancement of AI server platforms and the expansion of in-house AI accelerators by Big Tech companies.

In South Korea, SamsungElectroMechanics is accelerating the expansion of its supply of high-value-added MLCCs. This month, the company signed a 1.0722 trillion won supply contract with a major global corporation for MLCCs used in AI servers. This is the largest long-term supply contract for MLCCs in the company’s history, with deliveries scheduled for the entire year of 2027. SamsungElectroMechanics currently has long-term supply agreements (LTAs) with about 10 global customers and is pursuing additional contracts.

As leading manufacturers allocate a greater portion of their limited production capacity to high-value-added products such as those for AI, the MLCC market is expected to see accelerated differentiation by product type. While Korean and Japanese companies strengthen their dominance in the high-specification market, Chinese and Taiwanese companies are likely to increase their penetration in the general-purpose market, which is being relatively left vacant.

An industry official stated, “MLCCs for AI servers require high capacity and reliability, making it a market with higher profitability and higher technological barriers to entry compared to general-purpose products,” adding, “As investment in AI infrastructure continues, the trend among major manufacturers to focus their limited production capacity on high-value-added products will become even more pronounced.”