A price adjustment notice originating from the Kansai region of Japan has thrust MLCCs, often dubbed the "rice of the electronics industry," back into the spotlight. In the electronic component supplier circles around Huaqiangbei in Shenzhen, the reality of "haggling in the morning, prices rising by the afternoon" has shifted from a joke to a daily occurrence. Some scarce high-capacitance part numbers have increased over five-fold in price year-to-date, with distributor quotes fluctuating almost daily.

The price hikes are rooted at the manufacturer level. In March, Murata led the charge, raising prices on AI server and high-end automotive MLCC grades by 15%-35%. This was followed by Taiyo Yuden, which announced a price adjustment effective September 1st. More directly, Samsung Electro-Mechanics has uniformly raised shipment prices by 30% across its entire product line starting August 1st. The concurrent actions of these three global giants point decisively in one direction: capacity for high-end components is critically insufficient.

This, however, is not an industry-wide price surge. Inventory for consumer-grade MLCCs at the channel level remains higher than normal, with price increases limited to the 5%-10% range. In stark contrast, lead times for high-capacitance, high-voltage models used in AI servers have stretched to 12-16 weeks, with some extreme specifications scheduled out to the first half of 2027. While production capacity utilization at Japanese and Korean leaders stands at 90%-95%, their new capital expenditures are being directed entirely towards high-end multilayer production lines.

The industry's foundational logic has shifted. This isn't just a typical cyclical recovery; it's a permanent structural displacement in demand. As one industry insider noted, "In the future, only two types of MLCC companies will survive: those focused on the high-end AI/automotive segment, and those struggling in the low-end red ocean."

Back in China, A-share players in the MLCC sector, including Sunway Communication (300136.SZ), Fenghua Advanced Technology (000636.SZ), and CCTC (Three-ring Group) (300408.SZ), are vying for a seat at the high-end table, each employing a distinct strategy.

AI Rewrites the Value of Capacitors

To understand why MLCC has become so bifurcated, one must first grasp the impact AI servers have had on these small components. A traditional server uses around 2,000 MLCCs, primarily on 8-16 layer motherboards for basic filtering and decoupling, with each unit costing mere fractions of a cent.

However, as Nvidia's architecture evolved from Hopper to GB300 and on to the VR200 NVL72 rack, the power delivery architecture has been completely restructured. The number of capacitors in a single 8-GPU training server has surged to 20,000-40,000 units. A GB300 NVL72 rack requires about 440,000 units, while the VR200 NVL72 rack sees this number soar to 600,000, an increase of over 30% compared to the previous generation.

With AI chip power density nearing physical limits, "near-field power delivery" demands that capacitors operate reliably at 125°C. Capacitance values have jumped from the conventional 1-10μF to over 100μF, layers have doubled, and dielectric thickness is being pushed down to the micron level. Consequently, the price per unit has skyrocketed from 0.02-0.03 yuan to 0.5-2 yuan, representing a tens-to-hundreds fold price difference. The expanding new energy vehicle market is also driving significant demand, with a pure electric vehicle (BEV) using approximately 18,000 MLCCs—six times more than a traditional internal combustion engine vehicle. The 800V platform's demand for high-voltage, high-capacitance models has fully saturated automotive-grade production lines.

Murata's president, Norio Nakajima, had already predicted at an analyst meeting in May that robust demand for high-end MLCCs would persist at least until 2030. Moreover, Citi reports indicate that Murata plans to invest approximately 40 billion yen each in its 2026-2027 fiscal years to increase production. Crucially, the proportion of new capacity dedicated to high-end AI server components is slated to increase from 30% to over 45%. This signals that mid-to-low-end production lines will not be revived but will likely be further phased out.

Price data underscores the severity of this divide. Japan's MLCC export average price on a yen basis rose 16% year-on-year in April 2026, with export value up 28% (per Japan's Ministry of Finance trade statistics). At the contract level, manufacturers have raised prices 15%-35% for high-end AI and automotive models. In the spot market, certain scarce high-capacitance AI part numbers may have seen year-to-date increases that far exceed previous expectations. In contrast, price adjustments for ordinary consumer-grade products have generally been limited to 5%-10%.

A recent research report from GF Securities analyzed Japanese customs data, and after stripping out currency and CPI inflation effects, the actual average export price of Japanese MLCCs in the first half of 2026 was only 4.4% higher than in 2017. Current prices have merely returned to pre-2017 cycle levels, still well below the peaks of the last cycle. In other words, the price correction for high-end products has only just passed its first phase.

According to the "2026 MLCC Industry Research Report" (the "Report") recently released by JIWEI Consulting, the industry in 2026 is characterized by extreme divergence: "sharp increases in the high-end, slight increases in the consumer." This is indeed the reality. Contrasting with the fervor in the high-end market, consumer-grade MLCC is experiencing "passive price increases"—driven not by end-demand, but by the displacement effect as Japanese and Korean manufacturers shift their production capacity for consumer-grade X5R dielectrics towards X6S and X7R grades used in AI servers.

Spot prices for low-capacitance general-purpose models like 0402 and 0603 (under 1μF) have softened since Q2 2026, with some specifications down over 40% from their 2023 peaks. However, mid-to-high capacitance consumer grades like 22μF and 47μF have seen a surge in channel rush orders due to supply contraction, with some spot prices up 2-3 times since the beginning of the year. Taiyo Yuden took the lead in April 2026 by raising prices 6%-13% on consumer low-capacitance and automotive MLCCs. Samsung Electro-Mechanics is also considering following suit, prompting some distributors to engage in preemptive stockpiling. This has created a divergent pattern where ODM actual orders decline while channel orders climb. TrendForce anticipates that consumer-grade mid-to-high capacitance product quotes will continue to rise in Q4 2026, with an average increase of 10%-20%.

This is the "fire and ice" scenario of the MLCC world: General-purpose parts for consumer electronics cost pennies, while the ultra-high-capacitance, high-temperature-resistant special models used in AI racks can command up to 2 yuan per unit in the spot market, often facing shortages.

Multiple Paths to High-End Breakthrough

With Japanese and Korean supply locked down and high-end lead times exceeding 20 weeks, domestic Chinese MLCC manufacturers are charting distinct courses based on their strengths.

Sunway Communication is pursuing a path where a major downstream buyer internalizes its supply chain. On the evening of July 14th, the company announced plans to acquire a 55% stake in its associate company, Sunway Electronic Technology (Yiyang) Co., Ltd., for up to 1.1 billion yuan. Following the transaction, its stake will rise to 70% and the company will be consolidated. At the same time, all parties intend to increase capital by an additional 1 billion yuan. While the target's book value might not look impressive, Sunway Communication is securing a ticket to the high-end MLCC arena. The logic is clear: as a core supplier of Apple antennas, wireless charging, and RF front-end modules, high-end MLCCs are a critical upstream component. Price hikes from giants like Murata directly eat into its profits. By extending its supply chain upstream, it avoids being choked by original manufacturers.

Financially, Sunway Communication's H1 2026 results showed revenue of 4.032 billion yuan, an 8.88% year-on-year increase. Excluding non-recurring items, net profit was 141 million yuan, up 22.05% year-on-year, and surged 71.44% when excluding the impact of exchange gains/losses. The comprehensive gross margin improved to 19.99%, a 1.38-percentage-point increase. The ramp-up of high-end MLCC production is a key highlight in its semi-annual report and is poised to become its second growth curve. Currently, its subsidiary Yiyang Dianke has achieved stable mass production of multiple high-capacitance part numbers, passed certification from major domestic customers, and is supplying steadily. It has also begun its global expansion, with certain part numbers in qualification with North American key customers, while server/data center clients are also being developed.

While Sunway Communication might be a newcomer entering MLCC through its module customer base, Fenghua Advanced Technology, a veteran state-owned leader in the A-share MLCC market, is hitting its profit inflection point. In H1 2026, it reported revenue of 3.5 billion yuan, a 26.28% year-on-year increase, and net profit of 290 million yuan, up 74.01%. Q2 alone saw a sequential net profit increase of 127.8%, with both revenue and core production/sales hitting record highs.

The significance of this inflection point lies not just in the growth rate, but in the changing revenue mix. Its semi-annual report shows sales in the automotive electronics segment up 30% year-on-year, industrial control up 54%, and smart terminals up 14%. Sales of its new supercapacitor product line increased by 85%, with breakthroughs in emerging fields like AI computing, energy storage, and low-altitude economy. The rebound in average selling prices of its core products lifted the gross margin of its electronic components and materials business to 17.31%. Supporting this performance is a shift in product development from "PPT plans" to "specific, mass-producible specifications." While gaps remain compared to Murata and Samsung, Fenghua now has multiple high-capacitance MLCC models suitable for AI servers that have passed customer certifications and gained entry into supply chains—not just a mention in an annual report.

CCTC (Three-ring Group) perhaps comes closest to the "Chinese Murata" narrative as a vertically integrated player. The company delivered impressive results in the first half of 2026 too, with revenue of 6.423 billion yuan (up 54.82%) and net profit of 1.932 billion yuan (up 56.18%). Non-GAAP net profit reached 1.809 billion yuan, a 69.66% increase.

Its fundamental approach differs from the other two: CCTC self-supplies everything from ceramic powder and tape casting to lamination and co-firing. It has mass-produced palladium-silver electrode and ultra-thin dielectric technologies, and its lab-grade perovskite composite ceramic dielectric constant has surpassed 20,000. In July, it listed on the Hong Kong Stock Exchange to raise funds for overseas high-end capacity. Instead of chasing consumer-grade market share, CCTC is directly targeting the AI server high-capacitance/high-voltage and automotive segments, with a full line of products covering various AI server and NEV scenarios. In April 2026, it raised prices on high-end products by 20%-30%, becoming the first of the three Chinese companies to demonstrate high-end pricing power. The advantage of vertical integration is amplified during powder price upcycles—high-end powder accounts for 35%-45% of high-capacitance product costs. With tight supply for materials from companies like Sinocera and Sakai Chemical, CCTC's self-supply acts as a built-in buffer. The market values it for the "domestic substitution + high-end breakthrough" narrative, with its share price up roughly 151% this year. However, challenges remain: overseas client and server OEM qualifications are yet to be fully validated, vertical integration implies heavy asset investment, and the pace of overseas expansion carries uncertainty.

While these three companies take different paths, together they symbolize the Chinese MLCC industry's transition from "low-end substitution" to "high-end breakthroughs." With Murata scheduling orders into 2027 and Samsung Electro-Mechanics locking in major customers with long-term contracts, Chinese companies are finally finding their place at this high-end table.

The Underlying Threads of Powder and Equipment

The price hike letters are just the visible tip of the iceberg. Beneath the surface, three underlying supply chain segments—ceramic powder, nickel powder, and lamination equipment—are all under strain simultaneously.

Ceramic powder is the cost heart of high-capacitance models, accounting for 35%-45% of their cost. Five of the world's top seven powder manufacturers are located in Japan, with Sakai Chemical, Ferro, and Japan Chemical collectively holding over 60% market share. The particle size for high-end hydrothermally synthesized barium titanate must be controlled within 80-100nm, and achieving batch-to-batch consistency is an extremely challenging barrier. This means supply elasticity at the powder end is naturally very low; it's not something that can be quickly expanded just with capital.

An additional bottleneck arises from heavy rare earth elements. Dysprosium, terbium, and yttrium oxide are critical modifiers in high-end MLCC formulations, directly impacting the high-temperature stability of the ceramic dielectric. According to data in the Report, following tightened export controls at the end of 2025, Japanese manufacturers' rare earth inventory is expected to be depleted by June-September 2027, potentially forcing 20%-30% of capacity to be relocated. This places another constraint on an already tight powder supply, limiting the incremental capacity of overseas leaders who cannot secure sufficient modified additives.

The tighter the supply, the clearer the window for domestic substitution. Currently, Sinocera (Guoci Materials) has achieved full localization of MLCC ceramic dielectric powder, serving leading customers like Samsung Electro-Mechanics and Yageo. Bojun New Materials boasts globally leading ultra-fine nickel powder technology; its 80nm nickel powder has entered Samsung Electro-Mechanics' supply chain. Demand from AI servers for ultra-fine nickel powder directly contributed to its record revenue and profit in Q1 2026. The company plans to invest 124 million yuan to expand two nickel powder refinement production lines and has submitted an application for an H-share listing.

While overseas powder leaders are constrained by rare earth inventories and equipment delivery bottlenecks, domestic powder suppliers are gaining a valuable window to accelerate their entry into the high-end qualification cycle.

However, the bottleneck on the equipment front is even harder to circumvent than materials. The three most critical machines in an MLCC production line—the tape casting machine, lamination machine, and high-temperature co-firing furnace—are predominantly manufactured by Japanese companies, with delivery lead times ranging from 1 to 2 years. It's important to note that producing high-capacitance capacitors for AI servers isn't simply a matter of making a few more consumer-grade units. Consumer-grade MLCCs have thicker dielectric layers and fewer laminations, making each one very thin; a single square meter of film can yield hundreds of millions of units. In contrast, AI high-capacitance parts require dielectric layers under 1 micron and over a thousand laminations, making each unit several times thicker. This means the number of units that can be cut from the same square meter of film drops drastically to just tens of millions.

The Report provides a clear illustrative data point: converting a consumer-grade production line to produce AI ultra-high-capacitance models results in a capacity replacement ratio of 7:1. To free up capacity for 1 unit of AI high-capacitance production, 7 units of consumer-grade capacity must be sacrificed. Japanese and Korean majors are already running at around 95% utilization, with machines nearly fully operational. Without new equipment in place, they simply cannot increase output further.

Viewing these two underlying constraints alongside the capacity strategies of Japanese and Korean leaders clarifies the logic. Companies like Murata and Samsung Electro-Mechanics are proactively ceding mid-to-low-end capacity, channeling nearly all capital expenditure into high-end AI server and automotive multilayer lines. In essence, with a finite amount of high-end machinery and powder capacity, they cannot produce low-end products while also meeting high-end demand. When the forces of powder scarcity, equipment tightness, and leaders prioritizing high-end production converge, effective new supply for global high-end MLCCs before 2027 will be very limited. This suggests the upward shift in the average price of high-end MLCCs is not a short-term pulse, but a medium-to-long-term trend underpinned by material costs and equipment delivery cycles.

The true fortress of domestic substitution isn't just in "being able to make it." It lies in simultaneously overcoming three challenges: achieving self-sufficiency in high-end powder, securing lamination yield rates, and passing major customer certifications. The Report notes that mainland China imported over 2.56 trillion MLCC units in 2025. This massive import scale underscores the vast space for domestic players to leap from consumer-grade to high-end. Yet, to carve into the AI server and automotive mainline dominated by Japanese and Korean leaders, missing any of these three elements is not an option. This is the real competitive barrier beyond the visible price hikes: not just the size of the replacement space, but who ultimately secures the share in this round of high-end substitution.