The MLCC Market in Numbers

Pick any two research houses and you get two different answers for how big this market is, which is itself a planning lesson. For 2025, MarketsandMarkets, 2025 sizes global multilayer ceramic capacitor sales at $15.00 billion, rising to $21.93 billion by 2030 at a 7.9% CAGR. IMARC Group, 2025 says $11.9 billion, reaching $18.1 billion by 2034 at 4.62%. Expert Market Research, 2025 lands lowest at $10.70 billion with 5.00% growth to $17.43 billion by 2035, and Market Data Forecast, 2025 sits between them at $14.25 billion.

This is a five-company market

Concentration matters more to a new entrant than market size does. Murata, Samsung Electro-Mechanics, Taiyo Yuden, TDK and Yageo together hold roughly 68% to 72% of global revenue, according to the Passive Components Europe, 2025 summary of the segment. Murata alone is credited with 28% to 30%, and Samsung Electro-Mechanics with about 24% on first-half 2024 data. In the two most profitable corners the lead is sharper still: roughly 45% Murata and 40% Samsung Electro-Mechanics in AI-server MLCCs, and 40% to 50% Murata in automotive.

Chinese producers (Guangdong Fenghua, Sunlord, Three-Circle and others) have pushed their combined share to about 10%, mostly in commodity case sizes and consumer electronics. That is the segment where price discipline is weakest, and it is where a Western start-up with no scale should avoid pointing its plan.

Why demand is stretching supply in 2026

Count the parts. Sources compiled by Defiance ETFs, 2026 and Procurabl, 2026 put a smartphone at 1,000 to 1,200 MLCCs, a standard server at about 2,500, and an eight-GPU AI server at 15,000 to 25,000. A full NVIDIA GB200 NVL72 rack consumes roughly 440,000. Electric vehicles are quoted anywhere from 5,000 to 30,000 depending on the source and architecture, with 10,000 to 18,000 the most defensible planning range.

The market response has been unusually visible. Cosolvic, 2026 reports Murata raising prices 15% to 35% on AI-server and high-end automotive MLCCs in April 2026, its first broad increase in about three years. A Murata part used in AI accelerators (GRM31CR61A476KE15L) was quoted at $300 per 2,000 pieces in early April and had doubled to $600 by mid-June. Distributor lead times as of August 2026 run 14 to 18 weeks for standard parts, 15 to 20 weeks for high-capacitance or high-voltage parts, and up to about 40 weeks for AI-server grades. Murata added roughly ¥80 billion (about $500 million) of MLCC capex in April and says capacity is still insufficient, while its new Izumo factory does not reach full output until 2027.

Read that carefully before you write a market section. A shortage proves demand exists. It does not prove a new entrant can capture it, because the constrained parts are exactly the ones the incumbents have spent decades qualifying. A credible plan says which narrow slice of the shortage is open to you and why the incumbents are not serving it. The honest answers are usually one of: high-voltage and high-reliability parts in low volume, regional supply for customers who want a second source outside East Asia, or a distribution and design-in business that does not make the part at all.

Related reading: if your plan leans on automotive volumes, see our automotive PCB business plan template for how Tier 1 qualification timelines change the cash-flow forecast.

Four Ways to Enter the MLCC Business

"Multilayer ceramic capacitor business" covers four structurally different companies. They share a product and almost nothing else. Choose one before you touch a spreadsheet, because the capital, the margin and the time-to-revenue differ by orders of magnitude.

Capital figures for the fab models are derived from the plant-cost ranges in OpenPR, 2026; fabless and distributor ranges and all margin bands are Avvale planning estimates, not published statistics.

The fabless row deserves a second look in 2026. When a spot part doubles in two months, the contract fab keeps its own customers whole and the fabless brand sits at the back of the queue. The model works when you hold a prepaid capacity agreement, which means working capital, not just design capability. Lenders will ask to see that agreement. Without it, the plan reads as an inventory bet.

The specialty-fab row is where a funded founder with a materials or ceramics background has a real argument. Knowles Precision Devices (which absorbed Syfer in Norwich, Johanson, Novacap and Dielectric Laboratories) built a durable business on parts the Japanese and Korean giants ignore because the lots are too small. Nothing in the economics of that niche requires matching Murata's output.

What a Plant Costs and Who Funds It

The only published plant-level capex bands in the open literature are consultancy estimates, so treat them as scoping numbers. OpenPR, 2026 puts the minimum entry plant at about 500 million units a year for roughly $8M to $15M, mid-sized plants at $25M to $40M, and large facilities at 2 billion units a year above $60M. Divide it through and you get an implied $16M to $30M of capital per billion parts a year, remarkably stable across scale, which tells you that scale buys purchasing power and yield learning rather than cheaper machines.

Where the machinery money goes

An MLCC line is a sequence of ceramic and thin-film steps, and each step has a short list of qualified equipment vendors. For the sheet-forming stage, Hirano Tecseed (Japan, founded 1935) builds ceramic sheet forming lines specifically for MLCCs, inductors and thermistors. JTEKT Thermo Systems supplies MLCC and LTCC furnaces, and Tokyo Weld supplies the high-speed testing and taping machines at the back end, with more than 10,000 taping systems shipped for SMD chips. Chinese suppliers such as Uppershell, YSR and Dron sell complete lines at lower headline prices, with the usual trade-off of thinner process support and slower ramp to high layer counts.

• Slurry preparation and milling (bead mills, de-aeration): the dielectric particle size you can hold sets the thinnest layer you can sell
• Tape casting (slot-die or doctor-blade, in-line dryer, thickness gauge): the single most capital-intensive and yield-critical stage
• Electrode printing and stacking/lamination: layer count and registration accuracy decide capacitance per case size
• Cutting, binder burnout and sintering furnaces: reducing atmosphere for base-metal (nickel) electrodes, with tight temperature uniformity
• Termination, plating and end-of-line test: copper or silver termination, nickel and tin plating, 100% electrical test and taping
• Cleanroom, utilities and abatement: dry air, nitrogen, hydrogen blends, solvent recovery and exhaust treatment
• Reliability lab: HALT, temperature-humidity-bias, and thermal-cycling chambers, needed before any automotive or aerospace customer will audit you

Raw materials: a short supplier list

Two materials dominate the bill: barium titanate powder and nickel electrode paste. For barium titanate, the supplier names that recur in the market are Sakai Chemical, Nippon Chemical, Fuji Titanium, KYORITSU, Ferro and the Chinese producers Shandong Sinocera and Guangdong Fenghua (Verified Market Reports). For nickel powder and paste, TTI Market Eye notes that conversion is dominated almost exclusively by Japanese technology houses, with names such as Shoei Chemical, Sumitomo Metal Mining, JFE Mineral and Toho Titanium. The incumbents are also integrating backward: TDK and Nippon Chemical launched a joint venture specifically for MLCC materials.

Two planning consequences follow. First, a new fab buys from companies that also supply its competitors, and in a shortage the powder allocation goes to the largest accounts first. Your plan needs a qualified second source for each material. Second, dopant chemistry matters. China's April 2025 licensing controls on seven medium and heavy rare-earth elements, including yttrium and dysprosium (Tech Times, May 2026), still applied as of that report. Those elements are among the dopants used to tune X7R and X5R dielectrics, so supply-chain risk belongs in the risk section, not the appendix.

Funding routes by model

• SBA 7(a) and 504 (US): the SBA size standard for NAICS 334416 (capacitor, resistor, coil, transformer and other inductor manufacturing) is employee-based, roughly 500 to 550 employees depending on the table edition, so a start-up fab qualifies as small. The 504 program, which pairs a bank loan with a fixed-rate CDC debenture, suits the real-estate and equipment share of a $8M to $15M entry plant better than a 7(a) alone. Confirm the current figure in 13 CFR 121.201 before you cite it.
• Equipment finance and vendor credit: furnace and tape-casting vendors in Japan and China routinely offer staged-payment terms against delivery milestones. Model these as a 20% to 30% deposit and the balance across factory acceptance and site acceptance.
• Innovate UK (UK): funding is allocated competition by competition. UKRI announced £11.5 million across 16 projects for semiconductor manufacturing and supply-chain resilience, and a separate call offered up to £18.5 million for components for smart electronic platforms. These are collaborative R&D grants, not plant finance, but they de-risk the pilot line.
• Venture and strategic equity: MLCC fabs rarely fit a standard venture model because the payback is measured in years of yield learning. Strategic investors (a power-electronics OEM wanting a second source, a defence prime) are the realistic equity route. See our advanced ceramics business plan template for the strategic-investor framing.
• Customer prepayment: in a 40-week-lead-time market, a customer will sometimes fund tooling against a volume commitment. Put it in the plan only if you have a letter of intent.

For a fabless brand or a distributor, none of this applies. Working capital is the whole business, and the right lender is an asset-based line against receivables and inventory, not a term loan.

Process, Yield and Plant Operations

The product looks trivial: a rice-grain-sized ceramic block with two metal ends. The business is hard because a part rated 100 microfarads in a 1206 case can contain several hundred alternating dielectric and electrode layers, each a few microns thick. A pinhole, a delamination or a crack in any layer is a field failure. Yield is therefore the central variable of the financial model, and the plan should show it explicitly.

Three yield numbers belong in a bankable plan. Line yield is the share of green sheets that become sintered chips without gross defects; commodity incumbents run in the high 90s, and a new line will not. Test yield is the share of chips that meet capacitance, dissipation factor and insulation resistance after plating. Reliability yield is the share of lots that survive highly accelerated life testing and the customer's qualification protocol. A new entrant that assumes 95% across the board on day one will be marked down by every technical reviewer who reads the model.

Ramp realistically. A staged assumption most diligence teams accept for a greenfield specialty line: month 1 to 6 of production at 55% to 65% good-chip yield, month 7 to 12 at 70% to 78%, steady state in year 2 at 82% to 88% for specialty parts. Those are Avvale planning assumptions, not published figures, and your own pilot data should replace them as soon as you have any.

Staffing and shift pattern

An entry plant of around 500 million parts a year typically runs continuous operation because sintering furnaces are cheaper to keep hot than to cycle. Plan on 55 to 90 people at the entry scale: process engineers (ceramics, thin-film, electrical), line technicians on a four-crew rotation, a materials-and-quality group, and a small reliability lab team. Ceramics engineers are scarce; the main sources are the incumbents' alumni, Alfred University's glass and ceramics programme in the US, and Sheffield and Leeds materials departments in the UK. Budget for relocation and a retention bonus tied to the yield milestones.

Qualification: the clock that does not move

Industrial and consumer customers approve a new part in weeks. Automotive customers do not. The AEC-Q200 stress-test standard for passive components was revised to Rev. E in March 2023 and, once a part is approved, production is subject to a two-year qualification maintenance cycle (Passive Components Europe). The standard itself is only the entry ticket; the Tier 1 customer then runs its own audit, often a PPAP submission and a process audit before any volume order. Build 9 to 18 months between first sintered lot and first automotive revenue into the cash-flow forecast, and fund the plant through that gap.

Where MLCC Capacity Sits, and Why

Location is a bigger variable here than in most manufacturing businesses, because the powder and paste suppliers, the equipment vendors and the engineers are concentrated in a few places. The table below is the practical version for a founder choosing where to build or whom to contract.

Site list compiled from Microwave Journal, Knowles, KYOCERA AVX and TDK, 2022; qualitative commentary is Avvale's.

For a US plant, the geography that works is a mid-sized city near a ceramics or materials-engineering programme and industrial power at a reasonable rate: the Carolinas, upstate New York and Ohio are the usual shortlist because each has existing passive-component or advanced ceramics employers to hire from. For the UK, East Anglia (around the Knowles cluster) and the North East have both the engineering base and the grant eligibility. Whichever you choose, the plan should name the site, the utility capacity and the nearest qualified-technician pool.

Revenue Model and Unit Economics

MLCCs sell by the thousand or the reel, priced per part, and the price per part depends on four things: case size (0201 up to 2220), dielectric class (C0G/NP0, X7R, X5R, Y5V), capacitance and voltage rating, and qualification grade. A mainstream 0402 commodity part can sell for a fraction of a cent. A 1 kV C0G part in a 1812 case, qualified for medical or defence use, can sell for dollars. A plan that blends the two without saying so will not survive a technical read.

Price cycles are the main P&L risk

This industry swings between shortage and glut. The April 2026 price increases of 15% to 35% (Cosolvic, 2026) come after about three years without a broad increase, and the previous glut left commodity makers running below cost. A conservative base case assumes your average selling price in year 3 sits below the 2026 peak. A sensible sensitivity table runs ASP at -20%, base and +15%, with yield moving the other way in a downturn because fabs under price pressure tend to run lower-yielding lots for volume.

Worked example: a specialty high-voltage line (composite)

The numbers below describe a single specialty line, built for $9.5M of plant capex, selling high-voltage and high-Q parts into power conversion, medical and industrial customers. They are composite planning assumptions, not a quote from any real company.

Set that against the listed makers. Murata reported an operating margin of 19.4% for 2024 and in 2025 guided to operating profit of ¥280 billion on ¥1.8 trillion of sales, about 15.6%. Samsung Electro-Mechanics posted record 2025 revenue of KRW 11.3 trillion and operating profit of KRW 913.3 billion, about 8.1% across all its businesses (figures from Reportprime company profiles and the AsiaAI Samsung Electro-Mechanics profile). A 15.5% operating margin for a niche plant at scale is therefore in range, not heroic. What is aggressive is the path to get there.

The ramp

Total capital required to reach break-even on this ramp: $9.5M plant, about $1.8M of working capital, $4.7M of cumulative operating losses and roughly $0.95M of contingency (10% of capex), or about $17.0M. That total, not the $9.5M equipment line, is the number the funding ask needs to carry. The most common error in first-draft MLCC plans is stopping at the equipment quote.

Who buys, and how long it takes

• Distributors (the large franchise catalogue houses and regional specialists) take standard parts and carry stock. Quick to land, thin margin, and they will drop you when allocation tightens.
• Power-electronics and industrial OEMs buy high-voltage and high-capacitance parts for converters, drives and chargers. Design-in cycle 6 to 18 months.
• Medical and implantable device makers pay for traceability and low-volume lots; the qualification burden is lot-level documentation, not stress hours.
• Aerospace and defence primes require MIL-PRF-55681 or MIL-PRF-123 style qualification and, for US work, domestic-source assurance. Slow, sticky and high-margin.
• Automotive Tier 1s require AEC-Q200 and an audit, then reward you with multi-year programmes. Volume leader, longest payback.

The best plans pick two of these five and show a named pipeline, even if it is a list of design-in conversations. Pursuing all five from a 140-million-part line spreads engineering too thin.

Compliance, Tariffs and Export Controls

No licence is needed to make a capacitor. The regulatory work sits in four other places: environmental permitting for the plant, chemical-content rules on the product, trade rules on the way parts and materials cross borders, and customer-imposed quality standards that function as de facto law.

United States

• Classification: NAICS 334416 for SBA and statistical purposes; HTS 8532.24 for the finished ceramic dielectric multilayer capacitors on import and export.
• Air permitting (EPA / state agency): slurry solvents and binder burnout produce volatile organic compounds, so most plants need a state air permit, with the tier (minor source or major source) set by emissions. Expect 3 to 9 months and consultant fees of $20,000 to $80,000, an Avvale planning range rather than a published tariff.
• Worker exposure (OSHA): nickel and barium compounds both carry airborne permissible exposure limits, so the plan needs a ventilation and monitoring budget. Check the current OSHA tables for exact values.
• Chemicals (TSCA): if you import new ceramic powders or additives, confirm each is on the TSCA inventory before first shipment.
• Tariffs: Tariff Tracker, 2026 reports a 25% Section 301 rate on HTS 8532.24 goods from China, with 100electronics, 2026 describing a further 10% to 12.5% layer from 24 July 2026. Tariff law moves quickly, so verify the effective rate with a customs broker before pricing any imported chip or contract-fab part.
• Defence trade (if applicable): parts designed for military use may fall under ITAR or EAR controls, with registration at State or a licence classification at Commerce.

United Kingdom

• Environmental permit: from the Environment Agency (England), SEPA, NRW or NIEA depending on location, under the Environmental Permitting regime. Allow 4 to 12 months for a bespoke installation permit.
• UK RoHS and UK REACH: UK RoHS tracks the EU restricted-substances list; UK REACH requires registration of substances manufactured or imported above one tonne a year with the HSE.
• COSHH: nickel compounds trigger a COSHH assessment and workplace exposure limits under HSE guidance.
• Export controls: GOV.UK dual-use guidance lists capacitors under entry 3A201. The Knowles application note on dual-use legislation summarises the Nuclear Suppliers Group trigger: pulse-discharge capacitors above 1.4 kV, above 10 J of stored energy, above 0.5 µF and below 50 nH of series inductance. Most MLCCs fall below it, but a high-voltage specialty line may not, and exports then need an OGEL or a SIEL from the Export Control Joint Unit.

European Union and China

• EU RoHS (Directive 2011/65/EU): lead-free base-metal MLCCs sit comfortably inside the directive. The Commission's Delegated Directive (EU) 2025/2363 revised the lead exemptions in point 7(c), with revised 7(c)-I expiring on 30 June 2027 (TUV, 2025). A plan that sells any lead-bearing dielectric should model the sunset; one that does not should say so.
• EU REACH: an importer of record carries the registration duty for barium titanate and additives; a non-EU fab typically appoints an Only Representative.
• China (MOFCOM): licences are required for export of seven rare-earth elements including yttrium and dysprosium (Clark Hill). If any dopant or any contract fab sits behind that border, lead times are exposed to licensing queues outside your control.

Quality standards your customers will require

ISO 9001 is table stakes. IATF 16949 is what automotive buyers ask for, AS9100 is the aerospace equivalent, and ISO 13485 matters for implantable medical customers. AEC-Q200 Rev E (March 2023, now spanning 16 passive component categories, according to Passive Components Europe) governs automotive parts, and MIL-PRF-55681 and MIL-PRF-123 govern defence ceramic chip capacitors. Budget $60,000 to $180,000 and 9 to 15 months to reach IATF 16949 certification from a standing start, again a planning range from Avvale engagements rather than a published fee schedule.

MLCC Terms You Will Meet in Diligence

Technical reviewers on lender and investor panels will test your vocabulary. Use these correctly in the plan and the questions get easier.

• C0G / NP0: Class I dielectric with near-zero capacitance drift over temperature; low capacitance, high stability, used in RF and timing.
• X7R / X5R: Class II dielectrics with higher capacitance per volume and a defined drift band (plus or minus 15%) across temperature; the volume workhorses.
• DC bias derating: the loss of effective capacitance as DC voltage rises; a rated 10 µF part can deliver far less under bias, which is why AI-server boards use many parts per rail.
• BME / PME: base-metal versus precious-metal electrode systems. BME (nickel) is the cost-efficient mainstream; PME (palladium-silver) persists in niche high-reliability lines.
• Case size: imperial code such as 0402 (0.04 by 0.02 inch), 0603, 1206, 1812. Smaller is harder to make and more profitable per wafer-equivalent of material.
• Green sheet: the unfired ceramic tape on which electrodes are printed before stacking.
• Delamination and flex cracking: the two failure modes reviewers ask about first; one is a process defect, the other an assembly-stress risk.
• PPAP: the production part approval process that automotive customers use to accept a part and process.

Sample Business Plan Preview

Preview the structure and financial outputs a buyer receives. These mockups use the specialty-line example from the unit-economics section above.

What's in the Template

Every Avvale template is pre-structured for the industry. For a capacitor business, the sections are weighted toward the questions a technical reviewer asks:

• Executive Summary: the entry model, the slice of the shortage you can serve, and the total capital required
• Company Overview: legal structure, site, and the ceramics or applications background of the founding team
• Industry Analysis: market range, incumbent concentration, price-cycle history
• Customer Analysis: which two of the five buyer groups you target, with design-in pipeline
• Competitor Analysis: head-to-head against named incumbents in your niche, not the whole market
• Operations Plan: process flow, yield ramp, qualification milestones, supplier second-sourcing
• Marketing and Sales Plan: distributor strategy, application engineering, trade-show calendar
• Management Team: founders, process leads, advisory board, planned hires

The optional Financial Forecast add-on (included in our $300/£250 and $1,000/£800 packages) provides a 5-year Excel model with income statement, cash flow, balance sheet, break-even analysis, and startup capital requirements. For related structures see the advanced ceramics template and the free business plan templates hub.

Frequently Asked Questions