How AI Is Creating Electronic Component Shortages - Even for Parts That Have Nothing to Do With AI

Introduction

The narrative around AI and component shortages usually goes like this: data centres need GPUs, GPUs are in short supply, NVIDIA has a long waitlist. That part is true, but it is only a fraction of the story.

The real disruption is happening a few layers down the supply chain, in a place most procurement teams aren't looking: the foundries that make the everyday components your production lines depend on.

Microcontrollers. Analog ICs. Power management chips. MOSFETs. These are not AI components. They do not go into data centres. And yet the AI boom is making them progressively harder to source and in some categories, that pressure is building toward a significant shortage.

This article explains exactly why, what component families are most at risk, and what your procurement team should be doing right now.

Why the AI Boom Is a Problem for Components You Thought Were Safe

To understand the mechanism, you need to understand how semiconductor manufacturing capacity is allocated.

Chips are made in fabrication plants fabs that operate at different process nodes. Advanced nodes (3nm, 5nm, 7nm) produce the cutting-edge processors that power AI systems. Mature nodes (28nm, 40nm, 90nm and above) produce the workhorse components of industrial, automotive, and general electronics: MCUs, analog ICs, power devices, sensors.

For years, mature-node fabs provided a reliable supply of these components. They are less expensive to run, less technology-intensive, and have long served the steady, predictable demand of manufacturing industries.

Then AI arrived and the economics changed completely.

Advanced fabs TSMC, Samsung, and others have pivoted capacity aggressively toward AI accelerators, GPUs, and high-bandwidth memory. The margins are extraordinary: an H100 GPU sells for tens of thousands of dollars. A 40nm automotive MCU sells for perhaps €3–6. For a foundry trying to maximise returns, the choice is straightforward.

The consequence is a structural squeeze on mature-node capacity. As advanced-node fabs prioritise AI workloads, the overflow pressure customers who cannot get what they need from tier-one fabs cascades down to tier-two and tier-three foundries. Those foundries are now seeing utilisation rates climb sharply, even for components that have nothing to do with artificial intelligence.

The Components Most at Risk Right Now

Microcontrollers (MCUs)

MCUs are the nervous system of industrial equipment, automotive electronics and IoT devices. They are mature nodes 28nm to 90nm and have been historically stable, predictable commodities.

That is changing. Foundries are so focused on advanced AI chips that investment in older node capacity is declining, and the structural deficit this creates is starting to show. Industry analysts warned through 2025 that shortages of mature-node MCUs were building, with some 8-bit and 32-bit devices from major manufacturers already showing extended lead times. Several widely-used MCU families including devices from Microchip, STMicroelectronics, Infineon, and NXP appeared on end-of-life notices in 2025 as suppliers consolidated product lines to focus on higher-margin parts.

If your designs rely on specific MCU part numbers without qualified alternates, this is a risk you need to address now.

Analog ICs

Analog chips op-amps, ADCs, DACs, voltage regulators, sensors are the invisible backbone of virtually every piece of electronic equipment. They are also almost entirely produced on mature nodes.

The squeeze on mature-node capacity is hitting analog supply hard. Texas Instruments, Analog Devices, and other major analog suppliers have faced increasing pressure on their fab allocations as AI-related demand redirects capacity. Lead times in certain analog categories have extended significantly, and pricing has moved upward in a market that typically sees very gradual price changes.

Power Management ICs and MOSFETs

Power management is arguably the component category most directly caught in the crossfire. AI data centres consume enormous amounts of power Goldman Sachs estimated that AI infrastructure is dramatically reshaping power demand curves globally. The power components required to manage that load (PMICs, gate drivers, switching regulators, MOSFETs) are in acute demand.

But these same components are needed by every other industry: automotive, industrial automation, renewable energy, consumer electronics. The supply base cannot expand quickly enough to satisfy both AI-driven data centre build-outs and the rest of the world simultaneously.

An industry analysis from TrendForce published in early 2026 found that leading foundries had been reducing 8-inch capacity since the second half of 2025 but average 8-inch utilisation amongst the top 10 global foundries rose to nearly 90% as power management and power device demand from AI infrastructure ramped. The most exposed categories are products like PMICs and power discretes, which are still largely based on 8-inch processes.

DRAM and Memory (Indirect Impact)

While DRAM is not a traditional industrial component, its shortage illustrates the cascading mechanism clearly. HBM (High Bandwidth Memory) for AI accelerators is consuming an ever-larger share of wafer starts at memory fabs. This is displacing production of standard DDR4 and DDR5. Industry projections for 2025 showed DDR4 and DDR5 lead times exceeding 30 to 40 weeks in constrained categories, with multi-quarter price increases running as high as 40%. The phase-out of DDR4 is proceeding even while embedded systems and industrial equipment still depend on it a mismatch that will cause real pain for production teams sourcing these parts in 2025 and 2026.

The Structural Problem Is Worse Than It Looks

Here is the uncomfortable truth: the current situation is not a temporary allocation blip. It is a structural realignment.

The world is building more electronic products every year. Electric vehicles require dozens of MCUs per car. Industrial automation is expanding rapidly. Renewable energy infrastructure runs on power semiconductors. Medical devices, smart infrastructure, defence systems all of these depend heavily on mature-node components.

At the same time, the investment going into mature-node fab capacity is declining in relative terms. The economics are simply not attractive enough when advanced nodes command the attention of investors, governments, and foundry boardrooms alike.

In early 2026, industry analysis showed Chinese foundries leading the way with nearly 70% of new 12-inch mature-node capacity expansion, while Western and Taiwanese fabs continue with modest expansion. This concentration creates its own category of risk — a disruption to Chinese foundry output (geopolitical or otherwise) would have an outsized impact on global supply of these components.

S&P Global Mobility was direct in its assessment: another semiconductor shortage concentrated in mature nodes of 40nm and above is likely in the 2025–2026 window. Unlike the 2021 crisis which was driven by demand shock and pandemic disruption this one is structural, caused by deliberate underinvestment in unsexy but essential manufacturing capacity.

What This Means for Your Procurement Team

The companies that do well during this period are going to be the ones that move earliest. This is the way it works in practice.

1. Audit your BOM for mature-node dependency now

Map every MCU, analog IC, power device, and MOSFET in your bill of materials to its process node and its primary foundry. Identify where you have single-source dependency on constrained part families. This is not a one-time exercise — it needs to become a regular part of your procurement review.

2. Qualify second-source alternatives before you need them

The worst time to qualify an alternate component is when you already have your main part on allocation. “There is engineering and qualification work that needs to be done, and that takes time. Get started now, while supplies last, so you have a qualified backup when the primary source gets tight.

3. Extend your purchasing horizon

Standard 8-12 week purchasing cycles are no longer sufficient for components in high risk categories. Buying to 26–52 weeks of forecast coverage is a defensible strategy for MCUs, analog and power devices with known supply pressure.” The cost of holding a few months of additional inventory is trivial compared to a production line stoppage.

4. Work with a specialist sourcing partner

If a part goes on allocation or lead times exceed your buffer, you need access to a global independent sourcing network, not just the authorized distribution channel, which will be exhausted first. A specialist electronic component sourcing partner with verified global supplier relationships can source components that you cannot get through standard channels and often, a lot faster than you might expect.

5. Monitor EOL notices proactively

Suppliers are rationalising product lines to focus on higher margin parts with the result that component obsolescence is accelerating. In mid-2025 alone, last-time-buy notices were issued for several widely-used standard components from Texas Instruments, NXP, Analog Devices, Microchip, Infineon and STMicroelectronics. If you are not actively tracking EOL notifications for every part in your active designs, you are operating in a blind spot.

The Bottom Line

The AI boom is not just a story about GPUs and data centres. It is quietly reshaping the supply landscape for the everyday electronic components that keep manufacturing industries running.

It’s an indirect but powerful mechanism: AI demand pulls foundry investment into advanced nodes, starving mature-node capacity of capital and creating structural scarcity for MCUs, analog ICs, power devices, and other components the very components automotive, industrial, medical, and general electronics manufacturers depend on.

This is not a crisis yet. But the conditions for one are building, and the window for proactive action is closing.

The companies best positioned for the next 12–24 months will be the ones who have extended their supply horizons, qualified their alternates, and built relationships with sourcing partners who can navigate allocation conditions on their behalf.

How Globx Can Help

Globx is a supply chain specialist with over 12 years of experience sourcing electronic components for OEMs and EMS companies across Europe and worldwide.

When components go on allocation, lead times extend, or your standard distributor cannot fulfil we find them. Our global network of verified suppliers covers hard-to-find, allocated, and obsolete components across all major categories: MCUs, analog, power, discretes, memory, and passives.

AI-driven supply pressure services:

• Electronic component sourcing fast sourcing of allocated or hard-to-find parts from vetted global suppliers

• Solutions for component shortages Specialized support when your main supply chain fails

• BOM evaluation risk assessment of your bill of materials against current supply conditions, with alternate part recommendations

• Excess stock purchasing if you have over-purchased in anticipation of shortages, we buy your excess