Your line is tooled for a CPU-based AI server board, your DDR5 modules are landed, and the holdup on the bill of materials is a ceramic capacitor that cost cents a year ago. That scenario is now the realistic planning case for the first half of 2027, and the demand driver behind it is not GPU training this time.
According to TrendForce, in a publication dated October 29, 2026, the growth of inference applications and AI agents is generating a new wave of processor demand. Every processor needs power modules and DDR5 memory, and both of those sub-assemblies consume high-capacitance multilayer ceramic capacitors on the motherboard and on the memory module. The same TrendForce release states that the volume of high-capacitance MLCCs used around CPU motherboards and DDR5 modules has doubled, without specifying the baseline period. That caveat matters: the doubling applies to a specific zone of the board, not to global MLCC demand as a whole.
This article explains what is pulling supply, which platforms are competing for it, when the tightness peaks, and how OEM and EMS/CEM procurement teams can keep CPU and DDR5-dependent lines running through the window.
What Is Driving the High-Capacity MLCC Shortage Through 1H27?
The short answer: AI inference is multiplying the number of CPUs and DDR5 modules shipped into servers, and each of those units carries a dense population of high-capacitance MLCCs for power decoupling. TrendForce (October 29, 2026) describes this as a structural shift rather than a one-quarter spike.
A multilayer ceramic capacitor stores and releases charge to stabilize voltage rails. On a server motherboard, the capacitors sitting around the CPU socket and the voltage regulator stage are mostly high-capacitance parts in small case sizes, and these are the grades that are tight (see also: Samsung MLCC Lead Times Hit 61 Weeks: What Buyers Must Do Now). A DDR5 module adds its own power management IC and the decoupling capacitors that go with it. When TrendForce says processor demand is rising because of inference workloads, the mechanical consequence for the capacitor supply chain is more boards, more modules and more high-capacitance parts per rack.
TrendForce's MLCC Market Bulletin of October 19, 2026 frames it the same way: AI server demand is spreading from GPUs to CPUs and DDR5, sharply increasing high-capacitance MLCC usage, with high-end supply shortages expected to peak in early 2027.
Who is affected? Any OEM or contract manufacturer building AI server motherboards, CPU platforms or DDR5 memory modules, and by extension any industrial customer whose equipment shares those capacitor part numbers with the server market - including industrial equipment makers whose boards share part numbers with the server market (see also: Consumer MLCC Phase-Out: Why Your X5R Parts Face a 2029 Runway).
How Did the MLCC Market Shift From a GPU-Centric Cycle to a CPU and DDR5-Driven One?
The previous AI hardware cycle was dominated by GPU and high-bandwidth memory demand tied to model training. The current phase, according to TrendForce (2026), is driven by inference running on processors and server-grade DDR5, which redistributes capacitor demand across a different set of boards.
A KraneShares analysis published on September 1, 2026 explains why DDR5 is involved at all. AI servers rely on large quantities of server DDR5 for data preprocessing, model loading, scheduling and coordination between accelerators. Those are host-side tasks that run on the CPU and its attached memory, not on the accelerator. KraneShares cites an SK hynix statement that demand for high-density server DDR5 linked to AI inference rose 50% quarter-over-quarter. That figure is reported by KraneShares rather than established in a standalone regulatory or statistical filing, so treat it as directionally indicative rather than a planning constant (see also: DRAM Shortage 2026: Prices Up 444%, No Relief Before 2028).
The wider memory market gives context for how tight the whole stack is. KraneShares (2026) reports that Samsung and SK hynix raised 2026 contract prices for HBM3E by nearly 20%, and that HBM4 pricing is expected to land 55% to 70% above HBM3E. Those are memory figures, not capacitor figures, but they show that component tightness is not isolated to one family (see also: Memory Suppliers Lock 70% of Capacity: What It Means for OEMs).
The consequence for buyers is that demand is now diversified across processors, memory and passives at the same time. A contingency plan that covers a single component family will leave the line exposed on the others.
Which Platforms and Components Are Driving the 2027 MLCC Shortage?
Who Is Pulling High-Capacitance MLCC Supply?
Two platforms are named in the research as incremental consumers. TrendForce (October 29, 2026) expects shipments of the NVIDIA Vera Rubin platform to ramp month by month beginning in the fourth quarter of 2026, which means MLCC consumption from that platform grows progressively rather than arriving as a single step. The same TrendForce release, echoed by TokenPost, reports that AMD has started procuring specialized high-capacitance MLCCs for its MI450 platform.
Two major platforms drawing from the same high-end supply pool at the same time is the core of the problem. Neither is a substitute for the other, and neither relieves the other's demand (see also: Intel CPU Shortage 2026: 3 Bottlenecks Blocking Your Xeon Supply).
What Explains the Doubling of MLCC Volume Around CPUs and DDR5?
The doubling reported by TrendForce (2026) and relayed by Futunn concerns the volume of high-capacitance MLCCs used around CPU motherboards and DDR5 modules specifically. Higher power delivery to the processor requires more decoupling capacitance, and more memory channels per socket require more modules, each with its own capacitor population.
Procurement teams should keep this distinction when assessing exposure. A part number used on a consumer device in a commodity case size may be unaffected. A high-capacitance part in a small case size that also sits on a server motherboard is the one to watch.
When Will the Shortage Peak, and When Does It End?
Hardware manufacturers surveyed by TrendForce (October 29, 2026) expect the shortage of high-end MLCCs to be most severe in the first and second quarters of 2027, meaning January through June 2027. TrendForce's own forecast, in both the October 19 Bulletin and the October 29 release, is that high-end supply gaps persist into 2027 and run through the full first half of the year.
The research does not state an end date beyond 1H27. Treating the second half of 2027 as relief is an assumption, not a published forecast.
How Does This Affect Pricing Power in Negotiations?
TrendForce (2026) expects the persistence of the supply-demand gap to let MLCC manufacturers retain the upper hand in price negotiations through 1H27 (see also: MLCC Prices September 2026: Standard Down, AI-Grade Up 105%). No reliable global percentage for MLCC price increases appears in the available public sources, and any specific figure circulating in the market should be treated with caution until an independent publication confirms it.
The practical reading for a buyer: leverage sits with the manufacturer until mid-2027 at the earliest, so delaying a negotiation does not improve the position.
What Does the MLCC Shortage Mean for OEM and EMS Procurement Teams?
Direct exposure is crucial for motherboard and DDR5 module makers linked to CPU platforms. That's where TrendForce (2026) sees a huge jump in high-capacitance MLCC use. Teams working on Vera Rubin or MI450 are exposed: TrendForce (October 29, 2026) expects Vera Rubin shipments to ramp from the fourth quarter of 2026, and reports that AMD has already started procuring high-capacitance MLCCs for MI450, ahead of the Q1 - Q2 2027 peak that surveyed manufacturers anticipate.
The biggest risk here is operational before it gets to finances. If you have a CPU or DDR5 line that relies on just one MLCC source with no backup, you're in trouble when supplies get tight (see also: Semiconductor Supply Risk & OEM Production Continuity: Dual Sourcing Limits). If that line stops, it also stalls the customer's delivery, turning a component issue into a contract problem, with costs that stack up across the channels detailed in Electronic Component Allocation Production Line Stoppage Cost: 6 Channels to Add Up.
Since TrendForce (2026) expects MLCC makers to retain negotiating leverage through 1H27, delaying negotiations is unlikely to improve a buyer's position.
Franchised distribution may struggle to absorb a demand increase of this scale, which can push buyers who normally rely solely on franchised channels toward alternative sourcing.
How Should OEMs and EMS/CEM Teams Get Ready for the Q1-Q2 2027 Peak?
First, lock in allocations with key MLCC manufacturers now, before Vera Rubin starts ramping up in Q4 2026 (TrendForce, 2026). Securing allocations before the Q1 - Q2 2027 window that TrendForce (October 29, 2026) identifies as the most severe is a reasonable precaution.
Get secondary sources qualified ahead of time. Qualifying an alternate capacitor vendor means checking electrical standards, reviewing reliability data, and validating builds on actual boards. Starting qualification only in Q1 2027 risks overlapping with the period TrendForce (2026) expects to be the tightest. If the part is on a CPU or DDR5 module, the order is clear: qualify first, then negotiate.
Forecast separately by platform. Vera Rubin and MI450 demand independently (TrendForce, 2026; TokenPost, 2026). Combining both into one MLCC line item can hide how each program's demand grows, meaning a hiccup in one won't help the other.
Build inventory buffers wisely. Since manufacturers hold the upper hand price-wise through mid-2027 (TrendForce, 2026), keeping extra stock of high-capacitance parts for CPU and DDR5 lines might be worth it instead of risking delays due to price spikes. Don't stock every capacitor on the BOM unnecessarily; focus on specific types and sizes where you really face exposure.
Stay updated on the source. TrendForce's MLCC Market Bulletin of October 19, 2026 and its press release of October 29, 2026 are the primary public sources on peak timing. Procurement teams should keep an eye on these updates for the latest 2027 guidance instead of relying on a snapshot.
What Does Shortage Sourcing Look Like When Franchised Channels Are Blocked?
Here's a typical scenario based on past issues during these component shortages.
An EMS provider is putting together server motherboards for an OEM on a CPU platform. In January 2027, the franchised distributor tells them that a specific high-capacitance MLCC part is on allocation, with next delivery pushed past the OEM's shipment schedule. The EMS has four weeks of stock. They have identified an alternate part number, but it's not yet qualified for that board.
A shortage sourcing strategy, like the one used by Origin Sourcing, tackles the problem on parallel tracks. The first track focuses on finding the actual stock: confirmed numbers, the holder's name, a date code, and a storage location. The second track is about qualifying the source, which means checking the chain of custody back to the manufacturer, reviewing packaging against spec, and ensuring there's no gap in traceability. The third track involves deciding what to do when stock is available; some lots might come faster from sources with weak documentation, while others might be slower but have better traceability. For server programs audited by customers, it's usually better to go with the traceable option, even if it means a shorter buffer. On the flip side, if it's for engineering validation rather than a shipped product, you might take the faster but riskier option.
The key takeaway is that during a shortage, the decision isn't just about whether a part is available. It's about balancing availability, traceability, lead time, and risk. This needs to be done with input from the buying team, rather than just filling a purchase order under pressure.
FAQ
What's a high-capacity MLCC, and why does it matter for AI hardware?
A high-capacity multilayer ceramic capacitor helps store and manage electrical charge on motherboards and memory modules, keeping voltage stable for processors and memory. These components are essential for CPUs, power modules, and DDR5 modules, so as shipment volumes grow, demand for MLCCs rises too, says TrendForce (October 29, 2026).
Why is AI inference causing this shortage instead of AI training?
Training mostly uses GPUs and high-bandwidth memory, while inference relies heavily on CPUs and server-grade DDR5 for tasks like preprocessing and scheduling, according to KraneShares (September 1, 2026). Each of these components has a lot of high-capacitance MLCCs, which is why TrendForce (2026) links the current shortage to the inference cycle.
When will the high-capacity MLCC shortage hit its peak?
Manufacturers surveyed by TrendForce (October 29, 2026) expect the worst of the shortage during the first half of 2027, specifically from January to June, with supply gaps lasting through that time according to their October 19, 2026 Bulletin.
Which platforms are most impacted by the MLCC shortage?
TrendForce (2026) points to the NVIDIA Vera Rubin platform, expected to ramp up in shipments from the fourth quarter of 2026, and the AMD MI450 platform, which is now securing specialized high-capacitance MLCCs. Both of these utilize the same high-end supply, meaning demand from one won't ease the pressure on the other.
Should procurement teams anticipate rising MLCC prices, and by how much?
TrendForce (2026) thinks MLCC manufacturers will maintain pricing power into mid-2027 due to ongoing supply challenges. There's no solid percentage increase out there, so any specific figures should be backed up by independent sources before using them in budgets or negotiations.