AI data center demand drives tight MLCC supply; AI-grade high-capacitance products may enter a supercycle stronger than that of 2018

Overview

This report summarizes the MLCC expert session from Jefferies Tech & Industrial Expert Day, featuring Kim Wang, a former sales manager at Sunlord Electronics. It focuses on the impact of AI data center expansion on MLCC supply and demand, pricing, inventory, lead times, and the competitive landscape. The core conclusion is that the AI-grade high-capacitance MLCC shortage may be more severe than during the 2018 smartphone cycle and could last at least through 2027; shortages of low-capacitance MLCCs have also emerged, but their persistence depends more on the implementation of capacity expansion and OEM strategies.

Core views

First, AIDC deployment is driving demand for high-capacitance MLCCs, with usage per rack rising from approximately 443k units for GB300 to approximately 600k units for VR200, resulting in a severe shortage. Second, the shortage has spread from high-capacitance to low-capacitance products, driven jointly by capacity reallocation, channel stocking, and supplier shipment controls and price increases. Third, Japanese and South Korean manufacturers have learned from the 2018 cycle and are more cautious on capacity expansion, with annual capacity growth plans of approximately 10%; Chinese manufacturers are more aggressive, but remain constrained in high-capacitance AI-grade MLCCs by equipment lead times, yields, and process know-how. Fourth, supply of AI-grade high-capacitance MLCCs is primarily controlled by Murata and Samsung, which together account for approximately 80–90% of global AI server MLCC supply.

Analysis framework

The report primarily uses expert interviews and industry supply-demand tracking, combining AIDC server demand, MLCC usage per rack, inventory levels, supplier capacity expansion plans, price changes, and lead-time changes to assess the strength of the MLCC cycle. High- and low-capacitance MLCCs are evaluated separately to avoid equating the structural shortage of AI-grade products directly with a shortage across the entire product category.

Methodology notes

• Expert interviewExpert Day Takeaways

  Industry expert takeaways

  The report assesses changes in MLCC supply and demand, inventory, pricing, and the competitive landscape based on the industry experience of a former Sunlord Electronics sales manager.

• Supply-demand analysisSupply-Demand Balance

  MLCC supply-demand shortfall

  The report derives the severity of the shortage from AIDC demand growth, increased usage per rack, capacity shifts, the pace of expansion, and inventory levels.

• Cycle comparisonCycle Comparison

  Comparison with the 2018 MLCC cycle

  The report argues that the 2018 cycle was mainly driven by smartphone upgrades and OEM stocking, whereas the current cycle is supported by more sustainable AI data center demand; therefore, the AI-grade high-capacitance MLCC cycle could be stronger.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• AI-grade high-capacitance MLCC

  Core beneficiary category of AIDC expansion

  Strengths

  Demand is more sustainable, the shortage could be more severe than in 2018, and it may continue through 2027.

  Weaknesses

  Capacity expansion is difficult, while yield and process know-how remain bottlenecks.

  Comparison

  Compared with low-capacitance MLCCs, the shortage of AI-grade high-capacitance MLCCs is more structural and persistent.

  Risks

  If AI server deployment slows or new capacity comes online faster than expected, the persistence of price increases and shortages could weaken.

• Low-capacitance MLCC

  Beneficiary category from shortage spillover and channel stocking

  Strengths

  Short-term price elasticity is significant due to capacity shifts, stocking, and shipment controls.

  Weaknesses

  The persistence of the shortage is less certain than for high-capacitance products and depends more on capacity expansion and OEM procurement strategies.

  Comparison

  Low-capacitance products may see very large price increases, but their supply barriers are lower than those of AI-grade high-capacitance MLCCs.

  Risks

  If Chinese manufacturers and other suppliers bring expanded capacity online quickly, the shortfall could narrow.

• Murata / Samsung

  Leading global suppliers of AI server MLCCs

  Strengths

  Together they account for approximately 80–90% of global AI server MLCC supply, with notable technological and yield advantages.

  Weaknesses

  They are relatively cautious on capacity expansion, with limited capacity growth plans.

  Comparison

  They have greater competitiveness in high-capacitance AI-grade products than Chinese manufacturers.

  Risks

  Their share could be eroded if customers promote supply-chain diversification or Chinese manufacturers achieve breakthroughs in high-end processes.

• Fenghua / Three Circle and other Chinese suppliers

  Potential participants in domestic substitution and capacity expansion

  Strengths

  They are more aggressive in expansion; Fenghua has a broad product portfolio, while high-capacitance products account for approximately 70% of Three Circle's capacity and have already entered some domestic CSPs.

  Weaknesses

  High-capacitance AI-grade MLCCs remain constrained by 9–12-month lead times for Japanese equipment, low yields, and limitations in long-term process accumulation.

  Comparison

  Opportunities are more direct in low-capacitance and certain high-end products, but Chinese suppliers still lag Japanese and South Korean leaders in AI-grade high-capacitance products.

  Risks

  Expansion ramp-up, customer certification, and cost control may fall short of expectations.

Key data

• GB300 MLCC usage per rackApproximately 443k unitsUsed to illustrate the intensity of MLCC demand from AI server platforms.
• VR200 MLCC usage per rackApproximately 600k unitsFurther increases from GB300, reinforcing the AIDC demand driver.
• Taiwanese OEM inventoryApproximately 2.5 monthsThe report says inventory is at a historical low.
• Distributor inventoryApproximately 1.5 monthsLow inventory increases restocking and shortage risks.
• End-customer inventoryLess than one monthIndicates low downstream safety stock.
• OEM price increases from May to June20–30%Primarily targeting domestic distributors.
• Walsin Tech low-capacitance MLCC price increase100–600%Price increase for distributors on July 1.
• Potential low-capacitance MLCC shortfall90 billion unitsThe expert expects this could emerge as peak-season shortages intensify from September to October.
• Potential lead time18–24 weeksLow-capacitance MLCC lead times could be extended.
• Murata/Samsung AI server MLCC shareApproximately 80–90%Shows the dominance of Japanese and South Korean suppliers in AI-grade high-capacitance MLCCs.
• Annual capacity growth plans of Japanese and South Korean leadersApproximately 10%Reflects caution toward capacity expansion.
• Murata planned additional capacity10bn units/month by YE26Capacity expansion plan disclosed in the report.
• Samsung planned additional capacity5.1bn units/month by 2027Capacity expansion plan disclosed in the report.
• Fenghua planned additional capacity15bn unitsRepresents the more aggressive expansion of Chinese manufacturers.
• Samsung production-line conversion period2–3 monthsConverting consumer MLCC lines to high-capacitance products could result in approximately 20% capacity improvement.

Impact & implications

From an investment perspective, the report is more favorable toward AI-grade high-capacitance MLCC suppliers and manufacturers with high-end capacity, customer certifications, and accumulated process expertise. Price increases and longer lead times may improve short-term profitability, but they will also increase material costs and supply-chain risks for downstream server, consumer electronics, and EV customers. Chinese manufacturers may gain share opportunities through expansion in low-capacitance and certain high-end products, but they still face equipment, yield, and formulation-process barriers in AI-grade high-capacitance MLCCs.

Risks

• AIDC server deployment falls below expectations, weakening demand for high-capacitance MLCCs.
• Japanese, South Korean, or Chinese manufacturers expand capacity faster than expected, easing shortages and causing prices to decline.
• Low-capacitance MLCC shortages are more affected by stocking and supplier strategies, creating uncertainty over their persistence.
• The ramp-up of yields, equipment deliveries, and process know-how for high-end Chinese products may be slower than expected.
• If downstream OEMs adjust procurement strategies or destock, the price cycle could become more volatile.
• The report does not provide specific stock ratings or target prices; the investment conclusion requires further validation using company-level financials and valuations.

What to watch

• Whether the MLCC shortfall and lead times expand further during the September–October peak season as the expert expects.
• Channel and end-customer acceptance following price increases by Walsin Tech and other suppliers.
• Additional capacity from Murata and Samsung and the progress of Samsung's production-line conversions.
• Yield, certification, and customer adoption of high-capacitance MLCCs from Chinese suppliers such as Fenghua and Three Circle.
• Whether successive AI server platform iterations continue to drive MLCC usage per rack.
• Whether low-capacitance MLCC inventory levels recover from historical lows.