NVIDIA Corporation (NVDA.US) Rubin platform drives up MLCC value content; Morgan Stanley expects the market size to reach US$44.4 billion by 2031

Morgan Stanley's latest global technology industry report states that with the continued expansion of AI servers and data centers, multilayer ceramic capacitors (MLCCs) are becoming a new key beneficiary segment in the AI infrastructure supply chain. The bank believes that the investment opportunities brought by AI are spreading further from semiconductors such as GPUs to high-end electronic components such as MLCCs, and MLCCs are still in the early stage of the penetration cycle for AI servers and racks. Tightening supply and demand for high-capacity, high-value-added products is expected to drive changes in the industry's revenue structure, profitability, and valuation system. Morgan Stanley expects that, driven by continued growth in demand for small, high-capacity, high-value-added MLCCs from AI servers and data centers, the global MLCC market is expected to enter a period of marked acceleration in the coming years. The report estimates that global MLCC shipment value will reach US$44.45 billion by 2031, with a compound annual growth rate of 20.3% from 2025 to 2031; among this, AI servers and data centers will become one of the most important sources of incremental demand. Among major manufacturers, Morgan Stanley believes that global MLCC leader Murata Manufacturing and Samsung Electro-Mechanics (SEMCO) are expected to be the main beneficiaries of this trend, and continues to give both companies an "Overweight" rating; Yageo also maintains an "Overweight" rating, while Taiyo Yuden was upgraded from "Underweight" to "Equal-weight." AI investment opportunities spread from GPUs to MLCCs; the high-end market forms a Japan-Korea duopoly Morgan Stanley believes that the MLCC industry is undergoing an important shift in investment logic. In the past, the market focus of AI infrastructure investment was mainly on core semiconductors and computing hardware such as GPUs, CPUs, HBM, and networking equipment. However, as AI chip performance, server power consumption, and rack density continue to increase, requirements for power supply stability have also risen rapidly, and high-end MLCCs are becoming increasingly critical basic electronic components. MLCCs can be understood as miniature "energy storage reservoirs" in electronic circuits. GPUs, CPUs, and FPGAs switch repeatedly at nanosecond speeds and need to obtain large amounts of current instantly, while MLCCs can quickly release stored charge, filter noise, and stabilize the power supply voltage near the chip. As semiconductor processes continue to shrink, chips run faster while operating voltages decline, so even very slight voltage fluctuations can affect chip operation. Therefore, the higher the computing power and power consumption of AI servers, the stronger the demand for small, high-capacity MLCCs. More importantly, high-end AI MLCCs have high technical barriers. Morgan Stanley points out that high-capacity, low-ESL MLCCs used in AI accelerators require stacking hundreds or even more than 1,000 dielectric layers, while defect rates must be controlled at extremely low levels and strict customer certification must be passed. This makes the current high-end AI MLCC market highly concentrated. Murata and Samsung Electro-Mechanics together account for about 85% of the high-end AI MLCC market, forming a de facto Japan-Korea duopoly. Due to high entry barriers, Morgan Stanley believes that this round of supply-demand tightness differs from the traditional MLCC shortage cycle of 2017 to 2018. At that time, supply tightness attracted new entrants, and the market eventually rebalanced supply and demand quickly; by contrast, capacity expansion and certification for high-end AI products are significantly more difficult. At the same time, AI platforms are competing with smartphones, automobiles, and industrial equipment for advanced MLCC capacity and raw materials. Supply risk is gradually shifting from short-term shortages to long-term supply security issues, and long-term supply agreements between suppliers and customers are beginning to increase. Global MLCC market enters an acceleration phase; scale may reach US$44.45 billion by 2031 Morgan Stanley expects AI to significantly change the MLCC industry's previously relatively slow growth trajectory. The report shows that after consecutive contractions in 2022 and 2023, the global MLCC market returned to growth in 2024, with shipment value reaching US$14.67 billion in 2025, up 10.3% year over year. As demand from AI servers and data centers accelerates, the bank expects future growth to increase markedly, with global MLCC shipment value reaching US$44.45 billion by 2031 and a compound annual growth rate of 20.3% from 2025 to 2031. Among this, the growth rate of the AI server and data center MLCC market is even more astonishing. Morgan Stanley estimates that by 2031, MLCC shipment value used only in AI servers and data centers will reach US$23.33 billion, more than double its previous forecast of US$10.8 billion, corresponding to a compound annual growth rate of 57.1%, compared with the previous forecast of 38.1%. In other words, if this forecast materializes, by 2031 AI servers and data centers will contribute a considerable portion of global MLCC market demand, and the growth logic of the MLCC industry will increasingly shift from the traditional consumer electronics cycle to the AI infrastructure investment cycle. NVIDIA CorporationRubin platform brings a leap in demand; MLCC value per rack may surge 166% Morgan Stanley particularly emphasizes that NVIDIA Corporation's (NVDA.US) next-generation AI platform upgrade is significantly increasing the number and value of MLCCs required per server rack. The report estimates that compared with GB300 NVL72, MLCC demand for VR200 NVL72 racks will increase by about 80%. Among this, each GB300 NVL72 rack requires about 320,000 MLCCs, while the number required for VR200 NVL72 will jump to about 570,000. The increase in quantity is only part of it; more important is the upgrade in product specifications. According to estimates, the MLCC value per rack for GB300 NVL72 is about US$4,664, while VR200 will reach about US$12,411, equivalent to growth of about 166%. Among this, the MLCC value on the compute PCB is expected to rise from US$3,527 to US$9,239, while the switch board rises from US$515 to US$1,997. This means that AI server upgrades not only increase the number of MLCCs used, but also push the product mix toward higher-capacity, higher-unit-price products. Therefore, the revenue increment obtained by MLCC manufacturers may be significantly higher than simple volume growth. Morgan Stanley estimates that the proportion of high-capacity MLCCs above 47F used in VR200 will increase from 18% in GB300 to 31%. At the same time, the unit price of some smaller, higher-capacity MLCCs may reach 5 to 10 times that of traditional-specification products. The high-capacity MLCC segment above 47F alone could bring considerable new opportunities. Morgan Stanley estimates that such products could contribute 50%-60% of the new potential market size for cloud AI MLCCs in 2027, corresponding to about US$2 billion in additional market opportunity. In terms of volume, cloud AI demand for MLCCs above 47F may surge from about 4 billion units in 2025 to more than 40 billion units in 2027, growing more than 10 times in two years. This kind of demand growth may force major suppliers to reallocate capacity or even undertake large-scale expansion. Supply and demand for high-end AI MLCCs tighten; industry pricing environment may face a turning point While demand expands rapidly, the supply side cannot easily keep up. Morgan Stanley points out that the high-capacity MLCC market is still mainly concentrated among a small number of Japanese and Korean manufacturers, including Murata, Taiyo Yuden, and Samsung Electro-Mechanics. As demand for AI infrastructure accelerates further, capacity utilization for these products may tighten and may even squeeze MLCC capacity needed by other end markets such as automobiles, consumer electronics, and industrial equipment. The report also believes that the pricing environment in the MLCC industry may be at a turning point. Murata management recently said that rising production loads for high-end products are making it increasingly difficult for the company to fully meet demand for low-end products. Although Murata currently still insists on not adjusting prices based solely on short-term supply and demand conditions, Morgan Stanley believes the company is focusing more on high-end areas where it has competitive advantages and adopting pricing methods that can more fully cover large-scale capital expenditure. It is worth noting that Murata is relatively cautious about signing long-term fixed supply agreements for high-value-added MLCCs for AI and data centers. Morgan Stanley believes this may mean the company does not want to lock in prices now, because MLCC prices may still have room to rise further in the future. This is also an important reason why Morgan Stanley believes the value of this round of MLCC growth may be higher than traditional cyclical growth. As the product mix shifts toward high-margin AI products, manufacturers gain not only higher shipment volumes but also higher average selling prices and profitability, which is expected to form more sustained earnings growth and support higher valuations. Murata and Samsung Electro-Mechanics may be the biggest beneficiaries From the perspective of the competitive landscape, Morgan Stanley is most bullish on Murata and Samsung Electro-Mechanics. In the 2025 global MLCC market, Murata ranked first with a 40.8% market share, and Samsung Electro-Mechanics ranked second with 22.5%; Taiyo Yuden, TDK, and Yageo had shares of 11.3%, 6.9%, and 5.4%, respectively. Morgan Stanley believes that manufacturers able to supply small high-capacity MLCCs stably and at scale will gain the most, so Murata and Samsung Electro-Mechanics are in the most favorable position amid growing demand from AI servers and data centers. MLCCs themselves are also an important business for these manufacturers. In 2025, MLCCs accounted for about 50% of Murata's sales, 46% of Samsung Electro-Mechanics', 19% of Yageo's, and 70% of Taiyo Yuden's. Among them, Murata expects AI/data center MLCC sales to grow 100%-105% year over year in the fiscal year ending March 2027, with their share of the company's MLCC sales rising from the previous 10%-15% to 20%-25%. Morgan Stanley's own forecast is more optimistic, expecting Murata's sales in this business to grow about 120% over the same period and by another about 100% in the next fiscal year. Therefore, Morgan Stanley continues to list Murata as its top pick in the Japanese electronic components industry and maintains an "Overweight" rating. Share prices have fallen sharply from highs; Morgan Stanley believes conditions for repositioning are forming It is worth noting that MLCC-related stocks had already experienced sharp gains earlier this year. As of September 11, Murata, Samsung Electro-Mechanics, Yageo, and Taiyo Yuden had risen about 127%, 449%, 141%, and 160%, respectively, year to date. However, calculated from their respective highs in June to July, the four companies subsequently fell about 40%, 38%, 52%, and 59%, respectively. Morgan Stanley believes that considering demand for high-value-added MLCCs for AI servers and data centers will not only continue to grow in 2026 but is also expected to keep expanding after 2027, while earnings for major manufacturers are expected to continue rising, the recent marked correction from highs has made some stocks attractive again. Specifically, the bank maintains "Overweight" ratings on Murata, Samsung Electro-Mechanics, and Yageo, and upgrades Taiyo Yuden from "Underweight" to "Equal-weight." Based on the closing price on September 11, Morgan Stanley gives Murata a target price of JPY 11,000, implying about 50% upside from the share price of JPY 7,352; Samsung Electro-Mechanics a target price of KRW 2.62 million, implying about 87% upside from KRW 1.4 million; Yageo a target price of NT$1,050, implying about 93% upside from NT$544; and Taiyo Yuden a target price of JPY 10,500, implying about 14% upside. However, Morgan Stanley lowered Murata's target price to JPY 11,000 from JPY 12,500 this time, mainly because the risk-free rate used in its DCF valuation model was raised to 3.0% from 2.6%; Taiyo Yuden's target price was also lowered to JPY 10,500 from JPY 12,500. The bank believes that, compared with Murata and Samsung Electro-Mechanics, it may take longer for Taiyo Yuden to obtain significant benefits from growing MLCC demand from AI servers and data centers. Overall, Morgan Stanley believes that MLCCs are becoming another important link in the upstream diffusion of the AI capital expenditure wave, after GPUs, storage, networking, and power infrastructure. Its investment logic is no longer just a cyclical rebound brought by a recovery in traditional consumer electronics, but structural growth jointly driven by higher AI server power consumption, an increase in MLCC content per rack, product upgrades toward high-capacity specifications, and a concentrated high-end supply landscape. As NVIDIA Corporation's next-generation platform further increases MLCC usage and value per rack, Morgan Stanley expects high-end MLCC demand to continue expanding rapidly from 2027 onward. For leading manufacturers capable of producing small, high-capacity, high-reliability products at scale, AI infrastructure investment may not only bring shipment growth but also further amplify earnings elasticity through product mix upgrades and higher average selling prices. Murata and Samsung Electro-Mechanics are therefore viewed by the bank as the most direct beneficiaries of this round of MLCC growth cycle.