AI Server Demand Drives Across-the-Board MLCC Price Hikes
Claire Weston
Samsung Electro-Mechanics and Taiyo Yuden have announced MLCC price increases of up to 30% — driven not by raw-material costs but by a deliberate shift of capacity toward higher-margin AI servers, leaving less supply for everything else.
How hungry are AI servers for MLCCs?
MLCCs — multi-layer ceramic capacitors, the most common passive component on any circuit board — are used in far greater quantities inside AI servers than in conventional hardware.
Nvidia's GB200 server requires roughly 6,500 MLCCs per unit. The next-generation Vera Rubin platform needs about 12,000 — demand nearly doubles in a single generational leap.
This means → AI servers are not just "using a few more capacitors." They impose a step-change in both volume and quality, and the high-end parts carry margins well above consumer-grade.
Who is raising prices, and by how much?
Samsung Electro-Mechanics: MLCC prices up 30% from August 1, citing structural demand growth. The company says continued expansion and efficiency gains still cannot close the gap.
Taiyo Yuden: a further increase effective September 1, with warnings that delivery schedules may slip. Earlier this year it already raised consumer-electronics and select automotive MLCC prices by 6–13%, then added another round covering MLCCs, inductors, and ferrite beads.
In plain terms = both companies are saying the same thing: capacity can no longer keep up, and price is the result.
How bad is the order backlog?
The key metric is the book-to-bill ratio — new orders divided by shipments. A ratio above 1.0 means orders are piling up faster than product ships out.
As of late June: Samsung Electro-Mechanics 1.31, Murata 1.30, Taiyo Yuden 1.25 — all at the highest levels since the COVID-era shortage.
This means → for every 100 units shipped, 125–131 new orders are waiting. The backlog is still growing.
Why has Taiyo Yuden become the pricing bellwether?
The most aggressive price increases this cycle come not from market leader Murata or from Samsung Electro-Mechanics, but from the third-largest supplier, Taiyo Yuden.
This reflects a structural shift: as Murata and Samsung redirect more capacity toward AI-grade MLCCs, Taiyo Yuden's share of the mainstream market has grown — and with it, its leverage in price negotiations.
In plain terms = the top two went after the more profitable AI orders, leaving the third player as the biggest seller in the rest of the market — and in a stronger position to raise prices.
Are major buyers already locking in capacity?
Samsung Electro-Mechanics has signed two long-term AI-server MLCC supply agreements with hyperscale cloud providers, worth roughly KRW 454 billion and KRW 295 billion respectively.
This means → large buyers are shifting from spot procurement to forward capacity commitments — a classic signal of expected supply tightness.
What makes this price cycle different from past ones?
Previous MLCC price surges were typically triggered by raw-material inflation or cyclical shortages; the supply gap eased once new capacity came online.
The core difference this time: the capacity shift is deliberate and structural — manufacturers are choosing to allocate production lines to higher-margin AI products, not reacting to an unexpected shortfall.
In plain terms = supply is not broken; manufacturers are actively picking more profitable customers. Whether mainstream-product supply pressure eases depends on how fast new capacity actually comes online.
Content is for reference only, not financial advice.