AI Server Demand Pushes High-End MLCC Lead Times to 40 Weeks
Nashnova编辑部
AI servers need 5 to 13 times more high-end MLCCs than conventional servers, and Samsung Electro-Mechanics' lead times have stretched to roughly 40 weeks — supply-chain stress from the AI buildout is now reaching the humblest passive components.
What are MLCCs, and why do AI servers need so many?
MLCCs — multi-layer ceramic capacitors, fingernail-sized components that stabilize current and filter electrical noise — are among the most basic parts on any circuit board.
AI servers draw far more power through their GPUs, making power management far more complex. A single AI server requires 5 to 13 times more MLCCs than a conventional one, and demands higher capacitance, smaller form factors, and greater reliability.
This means → AI servers don't just "use more capacitors." They push the entire spec threshold up a tier; standard-grade parts simply won't work.
Lead times at 40 weeks — how tight is that?
Samsung Electro-Mechanics' (三星电机) average lead time on some high-capacitance MLCCs has reached roughly 40 weeks. Murata's (村田制作所) products at 1 microfarad and above stretched to 30 weeks in July, up from 24 weeks in June; some distributor quotes hit 36 weeks.
For contrast: Avnet Abacus's August guide puts mainstream chip ceramic capacitors at roughly 8 to 20-plus weeks, with an outlook rated "stable."
In plain terms = ordinary MLCCs are still readily available, but AI-server-grade parts now take the better part of a year. The market has split in two.
Why can't capacity expansion keep up?
Some capacity additions originally slated for Q4 2026 have been pushed back to 2027, further narrowing the window for near-term relief.
High-end MLCC manufacturing carries steep process barriers — more layers, thinner dielectrics, tighter yield requirements — so adding lines is not a quick fix.
This means → until new capacity comes online, the shortage of high-end MLCCs will persist for at least several more quarters; pricing and lead-time pressure won't ease soon.
What do Samsung Electro-Mechanics' and Murata's results reveal?
Samsung Electro-Mechanics' component division posted Q2 revenue of ₩1.6494 trillion, up 29% year-on-year and 17% quarter-on-quarter, driven mainly by AI server, networking, and power-application sales.
Murata's capacitor segment recorded ¥282.5 billion in revenue for fiscal Q1 (ending June), up 30% year-on-year. Company-wide orders hit a record ¥673.9 billion, lifting the book-to-bill ratio to 1.34.
Murata flagged rising long-lead-time orders and said it is monitoring whether customers are placing orders early out of supply-shortage fears. This reflects a shift toward "stockpile purchasing" downstream, which itself deepens the supply–demand imbalance.
What does this mean for buyers and investors?
Samsung Electro-Mechanics has signed long-term supply agreements with multiple major global tech customers; locking in capacity is becoming the industry norm.
In plain terms = the AI supply-chain bottleneck is no longer just GPUs and HBM — high-bandwidth memory, the fast-access chips mounted right next to GPUs — even the most basic capacitors are now a chokepoint.
This means → procurement teams need to secure high-end passive-component supply windows well in advance. For investors, MLCC leaders such as Samsung Electro-Mechanics and Murata are direct beneficiaries of the AI infrastructure spending wave — and this trend shows no sign of reversing in the near term.
Content is for reference only, not financial advice.