MLCC Price Surge 2026: How AI Demand Is Reshaping the Passive Component Market

In the sprawling ecosystem of electronic components, few parts are as ubiquitous — and as easily overlooked — as the multilayer ceramic capacitor (MLCC). Often called “the rice of the electronics industry,” these tiny components — typically just 2 mm long and 1 mm wide — stabilize voltage for virtually every chip in every device. A smartphone needs roughly 1,000 of them. An electric vehicle, over 10,000. And an AI server rack? Upwards of 600,000.

That last number is why MLCCs suddenly matter to everyone. In 2026, as AI infrastructure spending accelerates at an unprecedented pace, the passive component market is experiencing its most significant supply disruption in nearly a decade. Prices are rising across the board. Lead times are stretching. And the structural forces driving this shift are fundamentally different from the speculative cycle that roiled the market in 2018.

This article breaks down the data, the drivers, the competitive landscape, and the outlook — with actionable insights for engineering and procurement teams navigating this challenging environment.

1. The Price Surge: What’s Happening Right Now

The current MLCC price escalation began in late 2025 and accelerated sharply through the first half of 2026. Unlike previous cycles, which were often driven by channel speculation and hoarding, this round is rooted in genuine, structural demand growth — primarily from AI server builds.

Manufacturer-Led Price Increases

Manufacturer	Effective Date	Scope	Price Increase
Murata (Japan)	April 2026	AI server & automotive MLCC	+15% to +35%
Samsung Electro-Mechanics (Korea)	April 2026	Full MLCC series	+5% to +10%
Taiyo Yuden (Japan)	May 2026	Consumer & automotive MLCC	+6% to +13%
Walsin Tech (Taiwan)	June 2026	Chip resistors & select MLCC	+10% to +20%
Yageo / Kemet (Taiwan)	July 2026	All capacitor series (MLCC, Ta, Al, Film)	+10% to +30%

Beyond official manufacturer price adjustments, the spot market has seen far more dramatic moves. According to distributors in Shenzhen’s Huaqiangbei electronics market, popular high-capacity MLCC part numbers (such as Murata’s 1206-476, 47μF) saw prices climb from roughly 6–7 RMB per 1,000 pieces to 25–30 RMB — a four- to fivefold increase. Some ultra-high-capacity GPU-grade parts have channel prices 25x above the original factory price.

2. The AI Demand Engine: Why This Time Is Different

The single most important factor in the current MLCC supply crisis is the explosive growth of AI server deployments. This is not a speculative or inventory-driven cycle — it’s a fundamental, structural shift in demand driven by real end-use consumption.

MLCC Consumption per Server Platform

The math is striking. A standard enterprise server uses approximately 2,000 to 3,000 MLCCs. An AI server? About 28,000 — a 10-to-15-fold increase. And when you scale to a full rack, the numbers become extraordinary:

The picture becomes even more dramatic at the rack level. Morgan Stanley’s teardown of Nvidia’s next-generation VR200 NVL72 rack estimated the MLCC bill of materials at approximately $4,320 per rack — up 182% from the previous-generation GB300’s $1,530. Goldman Sachs now ranks MLCCs as the third-largest cost item in AI server BOMs, behind only GPUs and memory.

It’s not just quantity that’s surging — it’s also the specifications. The mainstream capacitance for AI-server MLCCs has jumped from 22μF to 47μF, and Nvidia’s Vera Rubin platform is now specifying 100μF parts at scale — a specification that barely existed in volume production before. The next generation will push toward 330μF. Each step up in capacitance requires dramatically more layers (from 100 layers to 500–1,300+), thinner ceramic films (down to 0.3–0.4 μm), and far lower yields.

3. Supply Constraints: Why Capacity Can’t Catch Up

On the supply side, the story is one of physics, chemistry, and time. MLCC manufacturing — particularly for high-capacity, high-reliability parts — is extraordinarily complex, and the industry’s ability to expand output is fundamentally limited.

The Capacity Math

Goldman Sachs estimates that the MLCC industry can expand capacity by only about 10–15% annually, while AI server MLCC demand is growing at 80% CAGR. This gap is structural, not cyclical.

Why Expansion Is So Hard

Several factors prevent rapid capacity expansion:

• Equipment lead times: Key manufacturing equipment — tape casting machines, sintering kilns, stacking machines — have lead times of 10 to 16 months. A greenfield factory from groundbreaking to first output takes at least 1.5 years.
• Yield penalty at scale: Standard consumer-grade MLCCs achieve 99%+ yields. High-layer-count AI-grade parts (500–1,300 layers) see yields drop to ~40%, with some ultra-small-form-factor parts as low as 10–15%.
• Capacity cannibalization: Producing one AI-grade high-cap MLCC consumes the manufacturing capacity of roughly four standard-grade parts. Even though AI servers account for only 2–3% of MLCC units, they already consume ~10% of total industry capacity — and this is projected to rise to 15–20% by 2027.
• Production cycle: The full 13-step manufacturing process for a standard MLCC takes about 27 days. For AI-grade ultra-high-cap parts, it exceeds 50 days.
• Material constraints: High-end MLCCs require ceramic powder particles under 80nm and nickel powder under 120nm — materials produced by only a handful of Japanese chemical companies (Sakai Chemical, Nippon Chemical).

“Customers’ inquiries for high-end MLCCs are double our current capacity. We simply cannot meet the demand. The tight supply situation for high-end MLCCs is likely to persist through this year and next.”

— Nakajima Norio, President, Murata Manufacturing (February 2026)

4. Competitive Landscape: Who Controls the Market

The global MLCC market is highly concentrated, particularly at the high end. Japanese and Korean manufacturers dominate, with Chinese firms rapidly building capability but still trailing in the most advanced specifications.

China’s Substitution Opportunity

As Japanese and Korean leaders redirect capacity toward AI servers, their spillover orders are flowing to Chinese manufacturers. The results are already visible in financial data:

Company	Q1 2026 Revenue	YoY Growth	Position
Sanhuan Group	¥2.68B	+46%	High-cap MLCC qualified for AI servers; 100% in-house ceramic powder
Fenghua High Tech	¥1.52B	+19%	Monthly capacity 500B+ units; AI server supply chain qualified
Boqian New Materials	¥410M	+64%	80nm nickel powder — one of few global suppliers for AI-grade MLCC

However, the gap in ultra-high-capacity products (100μF and above) remains significant. Industry sources estimate it will take 3–4 years for Chinese manufacturers to close the technology gap in the most advanced MLCC specifications, where the full process chain — from powder formulation to stacking to sintering — must achieve integrated mastery.

5. Raw Material Pressures

Beyond the demand-supply dynamics, raw material costs are adding fuel to the price fire. The key materials driving cost increases include:

• Silver: MLCC electrodes and pastes require silver and nickel. Silver prices have doubled over the past year, directly impacting electrode costs. Murata cited silver price increases as a primary justification for its April price adjustment.
• Nickel powder: Internal electrodes for high-end MLCCs require ultra-fine nickel powder (80–120nm) with 99.999% purity. Only a handful of companies globally — including Japan’s Boqian and China’s Boqian New Materials — can produce this at scale.
• Barium titanate (BaTiO3): The ceramic dielectric powder is the core material of MLCC. High-end parts require sub-80nm particles, dominated by Japanese chemical firms like Sakai Chemical and Nippon Chemical.
• PET release film: A critical consumable in the tape-casting process, with high-end variants (under 3μm) controlled by Japanese firms Toray and Lintec.

These upstream material dynamics mean that even when manufacturers want to expand capacity, they face bottlenecks not just in equipment, but in the fundamental inputs to the manufacturing process.

6. Impact Across Downstream Industries

The MLCC shortage is not affecting all sectors equally. The market is experiencing what analysts describe as a “structural split” — ice and fire:

Segment	Demand Status	Price Impact	Supply Outlook
AI Servers	Explosive growth	+30% to +50%	Tight through 2027–28
Automotive (EV/ADAS)	Strong, growing	+10% to +20%	Capacity being squeezed by AI
Consumer Electronics	Mixed, seasonal weakness	+10% to +20%	Tight near-term, may ease Q4
Industrial	Stable	+5% to +15%	Manageable

The key dynamic is capacity cannibalization: as manufacturers prioritize high-margin AI-server orders (where ASPs are 9–10x consumer-grade parts and gross margins can exceed 60%), they reduce supply to other segments. This means smartphones, TVs, and industrial equipment are effectively subsidizing AI server builds through reduced MLCC availability and higher prices.

7. Outlook: How Long Will This Last?

The consensus among major research houses is that this is a structural, long-duration cycle — not a short-term blip.

Key Forecasts

• Goldman Sachs: AI server MLCC demand to grow 4.3x from 2025 to 2030 (34% CAGR). The AI infrastructure investment peak may extend to 2030 due to power supply constraints.
• CICC (China International Capital Corporation): AI server MLCC demand to grow 87% in 2026 and 88% in 2027. Global server MLCC demand to grow 49% and 61% respectively.
• TrendForce: High-end MLCC supply tightness expected through 2027. Consumer-grade prices may stabilize in Q4 2026, but AI-grade parts will remain constrained.
• Southwest Securities: Peak supply-demand mismatch likely in H1 2027. Potential for structural overcapacity in mid-to-low-end products by mid-2028, while high-end remains in tight balance.

Murata has announced 800 billion JPY in additional capex for server-grade MLCC capacity, spread over two fiscal years. Samsung Electro-Mechanics is building a new factory in the Philippines (output expected in 2028 at the earliest) and has signed a record 1.5 trillion KRW long-term supply agreement. But even with these investments, capacity additions are projected at only 10–15% annually — far short of the 80% demand growth rate.

8. Strategic Recommendations for Procurement Teams

For engineering teams and procurement organizations navigating this market, we recommend the following strategies:

1. Secure long-term supply agreements now. Samsung Electro-Mechanics has confirmed it is moving from spot orders to binding multi-year contracts with AI customers. This trend will accelerate. If your volumes are significant, negotiate LTAs (long-term agreements) before the next round of price negotiations.
2. Diversify your approved vendor list. While Murata and Samsung EM dominate the high end, Chinese manufacturers like Fenghua and Sanhuan are rapidly qualifying for mid-to-high-tier applications. Approving second sources now — even if they’re not yet your primary supplier — creates optionality when supply tightens further.
3. Redesign for alternative specs where possible. If your application doesn’t strictly require 47μF or 100μF parts, consider whether 22μF or 10μF with additional paralleling can achieve equivalent performance. The price differential between high-cap and standard-cap parts is widening rapidly.
4. Build buffer stock for critical parts. Channel inventory is currently at healthy levels (~1.7 months for distributors, 2–3 weeks for end customers). This is far below the 6–7 months seen during the 2018 cycle. If you have critical MLCC part numbers, building a 3–6 month buffer is prudent — but avoid over-ordering, which would exacerbate the shortage.
5. Monitor the automotive vs. AI capacity shift. As Japanese and Korean manufacturers redirect automotive-grade capacity to AI servers, automotive MLCC supply is tightening. If your products serve both segments, plan for automotive-grade availability to worsen before it improves.
6. Engage with BOM consolidation specialists. Working with a sourcing partner who can aggregate demand across multiple customers gives you priority access to constrained supply. Torlong-Elec’s BOM consolidation services are specifically designed for this environment.

Conclusion: The “Next Memory”

Goldman Sachs has called MLCC “the next memory” — drawing a direct parallel to how DRAM and NAND became strategic chokepoints in the AI value chain. The comparison is apt. Like memory, MLCC demand is being driven by an irreversible secular trend (AI infrastructure buildout). Like memory, supply is concentrated among a few dominant players with high barriers to entry. And like memory, the gap between demand growth and supply capacity expansion is structural, not cyclical.

But there are important differences. MLCC technology iterations are slower than memory’s — the barrier is not lithography but materials science and process mastery. This means that once a manufacturer achieves a specification breakthrough, the advantage is more durable. It also means the current supply tightness, particularly at the high end, is likely to persist longer than the typical 12–18 month memory cycle.

For the electronics industry, the message is clear: the era of treating MLCCs as a commodity is over. In the age of AI, these tiny ceramic components have become strategic — and their pricing, availability, and sourcing strategy deserve the same attention that procurement teams give to GPUs and memory.

At Torlong-Elec, we’re helping our customers navigate this transition — from BOM analysis and alternative sourcing to long-term supply agreements. Contact us to discuss your specific needs.