The 2026 MLCC shortage is not really about network equipment — it is a side effect of AI servers pulling high-capacity MLCC capacity off the table. A single AI server cabinet burns more than 440,000 MLCCs, mostly high-capacity 0402 and 0603 parts, and that pulled the same lines that make the commodity grades used on every industrial poe switch and OLT board. The shortage then flows straight into switch and OLT BOM and lead times.
AI server cabinets use 440,000+ MLCCs, heavily high-cap 0402/0603; the top three makers tilt lines toward those profitable parts.
The high-end gap sits around 15–20% and only eases by late 2027; lead times stretched from 8–12 weeks to 20–26 weeks.
PoE and wide-temperature industrial models carry more MLCC in their power and filter circuits, so they feel the hit first.
An MLCC is smaller than a grain of rice, yet it is the "nail" on every network mainboard — decoupling, filtering, and regulation everywhere. In 2026 the shortage is not because the world ran out of MLCCs; it is because the specific general-purpose grades got "moved." AI servers take 440,000+ MLCCs per cabinet, largely high-capacity specifications, so the top three suppliers tilt production toward those high-margin parts and squeeze the mid- and low-end industrial and consumer capacity.
Note: The high-end gap holds at 15–20%, and a new production line takes 1.5–2 years from construction to volume output, with long lead times on the core equipment too. That is why most analysts expect the tightness to last until late 2027 — meaning more than a year of "tight balance" planning for procurement.
On a switch or OLT mainboard, MLCCs do the decoupling and filtering. A Rayin 4-fiber 8-electrical PoE industrial switch board commonly uses hundreds of them; an OLT line-card board, with higher port density and more power circuitry, uses no fewer. Each SFP port and its power stage adds decoupling MLCCs, so a dense line card with many SFP port slots accumulates the count quickly. MLCC is not the dominant share of the whole BOM — the active chips and DDR are bigger — but it is the cost leader among passive components.
| Dimension | Normal (2024–25) | 2026 tight period |
|---|---|---|
| High-end MLCC lead time | 8–12 weeks | 20–26 weeks |
| Industrial-grade price move | Stable | +300%–500% |
| MLCC count per unit | Hundreds to ~1,000 | Same count, hard to buy |
| Passive-part BOM impact | Baseline | Mid-single to low-double digits % |
Tip: An industrial poe switch and wide-temperature models have heavier power and filter networks, so their MLCC usage and price sensitivity are higher than a plain box switch. For any poe over ethernet deployment, the MLCC load scales with port count, so treat those models as the first to feel the pinch.
Lock orders early. With lead times at 20–26 weeks, forecast on a rolling quarterly basis and do not wait until the line needs parts.
Keep safety stock. Hold extra for industrial-grade and wide-temperature parts — those are the ones most easily squeezed by AI material.
Qualify multiple suppliers. Do not bet on a single brand; beyond Murata, Samsung, and TYK, keep domestic alternate channels to spread outage risk.
Design in redundancy. On new models, prefer interchangeable packages to avoid being locked by one scarce part number.
Warning: For makers exporting network equipment, shorten quote validity and write MLCC volatility into the commercial terms for long-lead orders. A quote priced at old MLCC levels can turn loss-making by the time you build.
How long will the MLCC shortage last? Most analysts expect the 15–20% high-end gap to hold until late 2027. New lines take 1.5–2 years to reach volume, so all of 2026 should be planned as a tight balance.
Why would AI servers hurt network-equipment components? An AI server cabinet uses 440,000+ MLCCs, mostly high-capacity grades. Big makers tilt lines toward those high-margin parts, squeezing the mid/low-end industrial grades and stretching their lead times.
How hard is the switch BOM hit? Roughly: when MLCC is over 60% of passive cost and passives are about 10% of BOM, the passive portion rises mid-single to low-double digits percent — and the terminal quote moves with it.
How much did Murata, Samsung, and TYK each raise? Murata lifted high-capacity parts +15–35% from March; TYK +10–30% from April; Samsung Electro-Mechanics about +30% across the board from August 1. Real conditions were often worse.
Are plain box switches less affected? Relatively, yes. PoE and wide-temperature industrial models have heavier power and filter circuits and more MLCC, so higher sensitivity — but with industry-wide lead times stretched, no one escapes the stocking pressure.
The 2026 MLCC shortage is a chain reaction of AI servers grabbing capacity: high-end lead times out to 20–26 weeks, the three majors raising prices in turn, and ultimately higher BOM and longer lead times for switches and OLTs. For any poe over ethernet rollout, fold the MLCC spike into the quote so margin holds. Procurement rides it out with early order locking, safety stock, and multi-source supply, while makers protect delivery through supply resilience.
About the author: Sara — Sara is a Customer Manager at Rayin with over 10 years of experience in the communications field. She specializes in technical product selection and writes guides and tutorials that help procurement engineers and system integrators solve problems more efficiently. In her free time, she enjoys badminton and swimming.
About Rayin: Shenzhen Rayin Technology Co., Ltd. — Company Profile
