AI drives MLCC boom: Murata actively expands production, Samsung Electro-Mechanics secures 1.5 trillion won order

```

Artificial intelligence is triggering the most intense round of supply and demand restructuring in the global MLCC (multilayer ceramic capacitor) industry in history. Goldman Sachs has characterized the current boom cycle as the "largest in scale and longest in duration" for this industry, as explosive demand from AI data center construction has pushed global lead times for popular products to as high as 24 weeks—more than double the normal level.

Samsung Electro-Mechanical has made particularly noteworthy moves recently. According to The Elec, the company has just signed a supply contract worth 1.5 trillion Korean won with a major global technology company, marking the largest single order in the company’s history, with relevant revenue expected to be recognized starting in 2027. At the same time, Samsung Electro-Mechanical announced plans to build a new factory in the Philippines to expand MLCC capacity for servers.

The predicament of industry leader Murata Manufacturing similarly reflects the overall market tension. According to Mirae Asset Securities, current global delivery lead times for hot-selling MLCCs have stretched from the normal level of about 10 weeks to as long as 24 weeks. Murata announced a capital expenditure of 250 billion yen (about 1.56 billion USD) for the 2026 fiscal year, including an additional 80 billion yen specifically for expanding server-level MLCC capacity, with current utilization rates nearing 95%.

This AI-driven supply and demand mismatch is transmitting to the price side. According to Citrini Research, spot and distributor prices for consumer-grade MLCCs have already risen by 20% to 40% compared to previous levels. Analysts expect that as industry contract prices are renegotiated quarterly, price increases will further spread to long-term customers, and investor expectations for supplier profitability are being rapidly upgraded.

AI Server Demand Surges, MLCC Enters Strongest Super Cycle Ever

Large-scale expansion of AI data centers is the core engine driving this MLCC super cycle. An AI server requires as many as 28,000 MLCCs, about 13 times more than a standard server. NVIDIA's next-generation computing platform Rubin architecture will increase MLCC usage per board from 6,500 in the current GB200 platform to 12,000.

Beyond AI demand, electric vehicles and humanoid robots are also continuously driving consumption. According to Murata Manufacturing, an electric vehicle equipped with Level 2+ autonomous driving features requires more than 10,000 MLCCs, while a smartphone only needs 800 to 1,000.

Citrini Research estimates annual growth in AI servers is about 80%, far outpacing any supplier’s capacity expansion rate and thereby creating a structural supply-demand gap. Taiyo Yuden CEO Katsuya Sase directly described this wave of demand as "terrifying."

Samsung Electro-Mechanical Deploys Silicon Capacitor Integration Strategy, Targeting AI Differentiation Track

As competition in traditional MLCC intensifies, Samsung Electro-Mechanical is betting on a more differentiated technical path—integrating silicon capacitors (Si-Cap) with MLCCs and FC-BGA packaging substrates into an all-in-one solution.

Silicon capacitors are passive components made using silicon wafers as the base material. Energy storage is achieved by etching tiny holes in the wafer and placing electrodes inside; device thickness can be compressed below 100 micrometers. This technology provides more stable voltage output in high power density chip packaging environments, a key feature needed by AI servers. Samsung Electro-Mechanical expects the silicon capacitor market to grow at an annual rate exceeding 18%, with applications extending from mobile devices to AI servers, automotive electronics, aerospace, and optical communications.

Kim Won-ki, head of Samsung Electro-Mechanical Si-Cap development team, said the company is incorporating silicon capacitors into its packaging substrate product portfolio as part of its comprehensive solution strategy, and has begun mass production of silicon capacitors embedded in FC-BGA substrates, which are widely used in AI data centers and PC processors. Reportedly, Samsung Electro-Mechanical has supplied silicon capacitors to Marvell Technology’s AI ASIC and Samsung Electronics’ Exynos 2600 application processor.

According to Maeil Business Newspaper’s industry analyst commentary, Samsung Electro-Mechanical is currently the only company with all three capabilities: MLCC, FC-BGA packaging substrates, and silicon capacitors, while Murata Manufacturing and TSMC also manufacture silicon capacitors but lack comprehensive coverage. Technologically, Samsung Electro-Mechanical uses a DRAM-based architecture, leveraging the fine process technology accumulated during DRAM linewidth scaling; TSMC uses a logic-process-derived trench structure. Samsung Electro-Mechanical conducts batch production on 300mm wafers, with wafer manufacturing outsourced to foundries, operating a fabless model.

Murata Commits Major Expansion, Still Difficult to Match AI Demand Growth Pace

As the world’s largest MLCC manufacturer, Murata Manufacturing is addressing capacity constraints with a 250 billion yen capital expenditure plan for fiscal 2026, including 80 billion yen for server-grade MLCC expansion. However, even with these investments, Murata expects capacity to expand by only just over 20% above baseline in the next two years, which still falls far behind AI demand growth.

Raw materials and upstream supply chains also pose risks. Nano-scale ceramic powder, the critical dielectric material for high-end MLCC production, requires extremely high purity and is dominated by Japanese chemical manufacturers like Sakai Chemical Industry and Japan Chemical Industry. Additionally, the main raw material for MLCC release films is crude oil, and heightened geopolitical tensions in the Middle East have significantly driven up related costs, further constraining manufacturers' cost controls.

For Samsung Electro-Mechanical, apart from the silicon capacitor strategy, reports indicate its new Philippine factory will begin construction this year—dedicated to expanding high-end MLCC capacity for AI servers.

Prices Continue Rising, Analysts Diverge in Views

Supply constraints are steadily flowing through to prices. According to Citrini Research, spot and distributor prices for consumer-grade MLCCs have already risen by 20% to 40%, partly driven by stockpiling and repeated ordering. Taiyo Yuden has notified customers about price increases for some products including MLCC, starting in May.

Murata Manufacturing hinted that if rising oil prices push raw material costs beyond internal absorption capabilities, they may consider raising contract prices. Morningstar director Kazunori Ito commented that an "extremely tight" supply situation could lead to price increases "beyond market expectations". Mirae Asset Securities analyst Junseo Park has raised his forecast for Samsung Electro-Mechanical's average MLCC price in 2027 by 10%, and noted that price hikes in some distributor channels have already been implemented. Given quarterly pricing negotiations, further price increases are likely to continue.

However, Bank of America Securities Japan analyst Masashi Kubota takes a more cautious stance. In a research report last week, he pointed out that some price increase forecasts based on supply tightening are "overly aggressive," and even if manufacturers push through price hikes, the premium mainly covers higher fixed costs, not improved margins.

Goldman Sachs: Cycle Still Early, May Continue Until 2030

The market’s biggest concern is how long this super cycle can last. Goldman Sachs judges that the industry is still in the early stages of this AI-driven cycle.

According to recent reports from Goldman Sachs analyst Daiki Takayama, Murata Manufacturing expects AI infrastructure investment to peak around 2028, but due to factors such as electricity supply shortages, the cycle could last until 2030. Morningstar’s Kazunori Ito expects AI server and infrastructure investment will "at least keep growing until 2028", providing fairly clear medium-term growth visibility for passive components such as MLCC.

Risk Warning and DisclaimerThe market carries risks and investment requires caution. This article does not constitute personal investment advice, nor does it take into account individual users' specific investment goals, financial situation, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article suit their particular circumstances. Invest at your own risk. ```