In Q2 2026, high-end AI and automotive MLCCs were tight and expensive, with lead times on high-end parts reported as long as ten months [S1][S2].
The market is splitting by specification: high-capacitance AI and automotive parts are tight, while many commodity low-capacitance grades remain closer to supply-and-demand balance [S1].
Why AI and EV demand are pulling MLCC capacity
Industry forecasts estimate that demand for high-capacitance MLCCs for AI servers will grow at 30% CAGR through 2030 [S1].
Electric vehicles create a separate requirement profile. Multi-inverter powertrains, onboard chargers and driver-assistance electronics require high-voltage and high-temperature capacitors that cannot be selected solely by capacitance and package size [S1]. The move toward 800 V DC architectures adds another qualification constraint because capacitor voltage rating, dielectric behavior and thermal performance must be matched to the bus and its transient conditions [S7][S6].
Industrial systems, telecom infrastructure and medical equipment also appear in the broader shortage-risk list, so AI and EVs are not the only sources of pressure [S4]. The distinction is technical rather than cosmetic: AI servers pull very high capacitance density into compact footprints, while vehicles add high-voltage, thermal and reliability requirements.
Volume growth and capacity concentration
MLCC demand tied to AI servers could rise 4.3 times between 2025 and 2030, with demand projected to increase by 10% to 15% annually [S3]. These are demand forecasts, not guaranteed unit shipments.
Supply is concentrated. Murata and Samsung together control approximately 84% of the high-end AI-server MLCC market, making allocation decisions at two groups influential across the data-center supply chain [S1].
That layer-count gap matters because capacitance density depends on dielectric thickness, layer count, electrode geometry and the manufacturing process, not simply on the case size printed in a datasheet [S1]. Buyers replacing a restricted high-end part with a nominal equivalent should treat capacitance, rated voltage, dielectric, ESR, insulation resistance, flexure and solder-process conditions as separate acceptance criteria.
Lead times reveal a K-shaped component market

Other high-end parts have been reported with lead times as long as 10 months [S2].
The market split is visible in pricing.
Samsung Electro-Mechanics started raising prices for the manufacturers it supplies directly, as well as for distributors [S2]. Taiyo Yuden implemented a second adjustment on September 1 covering more of its MLCC lineup [S2].
AI versus EV capacitors: selection criteria
AI-server MLCCs are selected for capacitance density, low equivalent series resistance, high-frequency behavior and compact placement around processors and power controllers. Their commercial exposure is concentrated in a small supplier group, and constrained part numbers have extended lead times of 20–26 weeks or more [S1].
Automotive MLCCs are selected around rated voltage, temperature, dielectric, mechanical robustness and AEC-Q200 qualification.
A practical comparison is therefore possible. AI grades score on capacitance density and high-frequency power delivery, automotive grades on voltage, temperature and reliability qualification, and commodity parts on price and broad availability. Lead time is the decisive near-term variable for constrained X6S and high-voltage devices, while technical compatibility remains non-negotiable [S1].
For system architects, the relevant passive chain includes power conversion, rack communications and remote monitoring. Specifications for an industrial router, protocol gateway or fiber converter still require the same component-level review, since upstream MLCC availability can affect server or network deployment schedules even when the capacitor sits inside a supplier subsystem.
Price pressure is spreading beyond AI grades

Initial price increases began with scarce AI-grade products, but by summer 2026 they had spread into ordinary commodity parts [S2]. Walsin Technology raised prices for some MLCCs on June 1, Yageo increased capacitor prices broadly on July 1, and Walsin raised consumer-MLCC prices by 5%–15% on the same date [S2].
Raw-material exposure remains a cost driver. Taiyo Yuden cited higher prices for barium titanate powder, nickel powder and rare-earth additives when its September 1 adjustment took effect [S2]. Manufacturers can therefore face pressure from both constrained qualified production capacity and input costs, although a manufacturer announcement does not prove that every part number received the same adjustment.
Buyer behavior can amplify the signal. The industry spent roughly four years, from 2022 into early 2026, reducing excess inventory, leaving distributors with little stock when AI-server and EV orders strengthened [S2]. Distributor restocking then competed with original-equipment demand, and quotes in the Shenzhen Huaqiangbei market were changing as frequently as every 30 minutes by late June [S2].
Procurement must separate scarcity from equivalence
High-capacitance shortages should be managed at the exact manufacturer-part-number level rather than by accepting any X5R or X6R substitute. Required checks include capacitance tolerance, rated voltage, dielectric temperature characteristics, DC-bias behavior, ESR, insulation resistance, package dimensions and termination layout [S1][S8].
For EV programs, add 800 V bus exposure, operating temperature, vibration and automotive qualification to the review [S6][S7]. For AI equipment, add capacitance density, placement footprint, power-network topology and the supplier’s documented high-end capacity allocation [S1]. A stock-status label without those parameters is inadequate evidence of interchangeability.
Commercial planning also needs multiple data points. Track manufacturer lead time, authorized-distributor stock, price-change notices, allocation status and the planned production volume of the approved device. Prices rose across the MLCC market in 2026, beginning with the scarcest AI-grade parts and spreading to ordinary commodity parts over the summer [S2].
Limits of the 2026 forecast

The available figures combine third-party estimates with supplier expectations, so they should guide scenarios rather than serve as guaranteed market outcomes. The 72.6 billion-piece 2026 forecast and 136.7 billion-piece 2027 forecast describe AI-server demand, not total MLCC shipments [S2]. MLCC demand tied to AI servers could rise 4.3 times between 2025 and 2030, with demand projected to increase by 10% to 15% annually [S3].
Lead-time evidence is more operational than market-size estimates. The extension from 8–10 weeks to 20–26 weeks or more applies particularly to selected AI and automotive codes, not every capacitor [S1]. Likewise, spot prices of two to three times prior levels reflect short-term channel conditions and should not be extrapolated across all grades [S2].
The clearest decision rule is straightforward: treat high-end AI, automotive and 800 V EV parts as allocated commodities, but approve substitutions only after electrical, thermal, mechanical and qualification review. Reserve capacity for qualified designs, monitor manufacturer notices and authorized-channel stock, and maintain a second source only after the alternative has passed those checks.