The Chip Shortage in 2026: What’s Still Constrained and How to Source Around It

If you’ve priced a DRAM line item recently, or tried to quote a Nexperia MOSFET with any confidence, you already know the answer: the chip shortage hasn’t gone away. It has changed shape. The broad, everything-is-scarce crisis of 2021–2022 gave way to a period of relative normalization, and now the market has tightened again — but in a more targeted and, in some ways, harder-to-plan-around way. In 2026, the shortages that matter are concentrated: high-bandwidth memory (HBM) and conventional DRAM, automotive-grade legacy discretes disrupted by the Nexperia geopolitical standoff, high-capacitance MLCCs, and advanced power semiconductors for AI and EV applications. Meanwhile, standard logic ICs, connectors, sensors, and mature analog parts remain largely available on normal terms.

This article covers where the pressure actually sits in 2026, what’s behind it, and — practically — what procurement teams and design engineers can do about it. The goal is a factual reference you can use for planning, not a summary of headlines you’ve already seen.

1. Is There Still a Chip Shortage in 2026?

Yes — but the better question is where. Industry forecasts have the global semiconductor market growing strongly in 2026, with the growth driven almost entirely by AI infrastructure spend. Inside that growth number, the supply picture is sharply uneven. DRAM contract prices rose steeply quarter over quarter through the first half of 2026, according to TrendForce’s quarterly pricing reports, and spot prices for high-capacitance MLCCs moved sharply upward over the same period. Lead times for the most constrained categories — HBM, automotive-grade SiC MOSFETs, and high-density passives — are being quoted in quarters rather than weeks; get current figures for your own part numbers from the manufacturer or distributor, because they move month to month.

At the same time, a separate and largely unrelated shock hit the market in late 2025, when geopolitical intervention in the operations of Nexperia — one of the world’s largest suppliers of legacy transistors, diodes, and MOSFETs — disrupted automotive supply chains across Europe, North America, and Asia. The automotive requalification process for alternate suppliers can take months, so the effects of that disruption have continued well into 2026.

The important point for procurement planning: this is a selective shortage, not an undifferentiated one. Your exposure depends entirely on what’s in your BOM. Standard commodity parts are sourceable. The categories described below are not — at least not on ordinary terms.

2. What Is Driving the 2026 Shortage

2.1 AI and the Memory Reallocation Problem

The root cause of the memory shortage is structural, not cyclical. Samsung, SK Hynix, and Micron — the three suppliers that between them make virtually all of the world’s DRAM — have systematically reallocated wafer capacity toward HBM and server-grade DDR5 to serve AI data center demand. Producing a gigabyte of HBM consumes several times the wafer capacity of standard DDR5, creating a genuine zero-sum allocation problem: every wafer committed to an HBM stack for a GPU is a wafer withheld from a laptop, router, or consumer-grade storage product.

By mid-2026, Samsung’s memory chief had warned publicly that significant shortages across memory products are expected to continue through at least 2027. SK Hynix’s CEO went further, cautioning that demand will exceed supply well into the next decade, describing a gradual easing rather than any sharp correction. New fab capacity from Micron and SK Hynix is not expected to reach volume production before 2027–2028 at the earliest, meaning the structural undersupply condition does not have a near-term resolution. IDC has described the situation as a “permanent reallocation” of manufacturing capacity toward AI rather than a standard inventory cycle that will self-correct.

For conventional DRAM and NAND buyers, the practical consequence is that allocation status matters more than willingness to pay. HBM is effectively allocation-only, and some mainstream DDR4 lines have moved to allocation terms with lead times quoted in quarters. Quotes that previously held for a full quarter may now be valid only for a matter of weeks.

2.2 The Nexperia Crisis and Its Ongoing Fallout

In late September 2025, the Dutch government invoked the Goods Availability Act to assume control of Nexperia, a Netherlands-based but Chinese-owned semiconductor manufacturer, citing national security and technology-transfer concerns. China’s Ministry of Commerce responded within days by imposing export controls that blocked Nexperia’s Chinese operations — including its primary packaging and testing facility in Dongguan — from exporting finished components to international customers. Nexperia is one of the largest suppliers of automotive-grade discrete semiconductors — the transistors, diodes, and MOSFETs embedded throughout vehicle electronics.

The disruption was acute. Automakers including Honda, Volkswagen, and Nissan warned of imminent production halts, with existing inventories reported at the time to be measured in weeks. A US–China trade truce in late 2025 technically restored shipments, but the damage did not disappear with the announcement. Automotive-grade components require AEC-Q100/Q101 qualification — a process that cannot be compressed into a few weeks — meaning buyers who needed a genuine second source were still working through requalification well into 2026. The German automotive industry association and the Alliance for Automotive Innovation both publicly stated that supply risks remained elevated for months after the truce.

The broader lesson the Nexperia crisis surfaced — that seemingly ordinary discrete components can carry significant geopolitical weight — has prompted European and US OEMs to reassess supplier concentration risk and to build continuous lifecycle-monitoring programs that can detect EOL and allocation risk earlier than a reactive scramble allows.

2.3 MLCCs and Power Semiconductors Under Dual Pressure

Multilayer ceramic capacitors and power semiconductors are being hit from two directions simultaneously: AI infrastructure and automotive electrification, both of which require these components in large volumes and at higher specifications than previous demand cycles. A single AI server rack can contain hundreds of thousands of MLCCs; a battery electric vehicle requires roughly five times as many MLCCs as a comparable combustion vehicle. Murata, Taiyo Yuden, and Samsung Electro-Mechanics — the suppliers that dominate high-end MLCC supply — have announced price increases, and lead times for high-capacitance, low-ESL parts have extended well beyond normal. The passive component market is broadly expected to remain constrained through at least mid-2027.

The picture for power semiconductors is more nuanced. There is a meaningful divergence between 6-inch automotive-grade SiC, where oversupply from earlier investment cycles has softened pricing, and advanced 8-inch SiC modules required for high-power AI data center applications, where capacity constraints and yield challenges during the wafer-size transition are driving significant tightening. Lead times for automotive-qualified SiC MOSFETs are being quoted in quarters rather than weeks, and power management ICs, IGBTs, and MOSFETs more broadly are extended across the industry.

2.4 Trade Policy as a Supply Chain Variable

Layered over all of the above is an ongoing pattern of US-China trade friction that shows no sign of becoming predictable. Tariffs on semiconductors, substrates, and assembly equipment have moved from theoretical to landing directly on purchase orders. Rare earth and cobalt export restrictions are adding material cost increases in some categories. The “country of diffusion” rules introduced under recent tariff frameworks add compliance complexity for any BOM with cross-border supplier dependencies. The practical implication for procurement teams: policy risk now needs to be tracked with the same discipline as price and lead-time risk. A component can move from readily available to effectively restricted in a matter of weeks when a geopolitical trigger occurs.

3. Category-by-Category Breakdown: What’s Constrained Right Now

The table below reflects mid-2026 conditions across major component categories. It is a planning reference, not a real-time quote — lead times and pricing are moving targets in this market.

One category distinction worth flagging for design teams: automotive-grade 6-inch SiC components have actually seen pricing soften due to earlier oversupply, while 8-inch AI-grade SiC modules — a separate product line for a different application — are among the tightest in the market. Grouping all SiC as either “fine” or “not fine” leads to the wrong sourcing decisions. Similarly, not all DRAM is equally constrained: enterprise-grade server DDR5 is tighter than consumer modules, and the spec you’re designing around matters for lead-time forecasting.

4. Sourcing Strategies That Work in This Environment

The teams navigating this market well share a few common practices. None of them are exotic — they are discipline and timing, applied before a shortage becomes a production problem rather than after.

Audit the BOM for Concentration Risk, Not Just for Price

The first step is knowing which line items are genuinely exposed. That means flagging every part sourced primarily from Nexperia, from the Samsung/SK Hynix/Micron memory allocation pool, from single-source SiC suppliers, or from manufacturers with recent EOL announcements. A BOM that looks fine on cost review can have three or four critically exposed parts that don’t surface until a purchase order comes back with a lead time measured in quarters. Tools that check lifecycle status, supplier concentration, and alternative part availability across the full BOM — rather than part by part — are significantly more efficient for this kind of risk audit. A BOM upload tool that returns availability and lifecycle status line by line supports this kind of multi-line review, and WIN SOURCE’s shortage and obsolescence management services are structured around exactly this kind of proactive lifecycle assessment.

Qualify Alternate Parts Before You Need Them

This is the most time-sensitive action on this list and the one most often deferred until it’s too late. For automotive-grade components, requalification under AEC-Q100 or AEC-Q101 can take months regardless of how motivated the engineering team is. For industrial or consumer designs, a pin-compatible alternate on a different supplier’s fab can often be validated faster, but it still needs engineering time and a qualification cycle. The right time to qualify an alternate is when the primary source is healthy and available, not when it’s on allocation. Alternative solutions for hard-to-source components are a core part of managing BOM risk in this environment, and identifying those alternatives during design or early production — rather than during a shortage — is where the real value is. Final approval of any alternate belongs to your own engineering team; a parametric match is necessary but not sufficient.

Move from JIT to Forecast-Based Ordering for Exposed Categories

Just-in-time manufacturing was always a trade-off between working capital efficiency and supply chain resilience. In an allocation-driven market, it’s not a viable strategy for the constrained categories above. Suppliers on allocation are prioritizing customers who share a rolling demand forecast, even an informal one. Buyers who show up with a spot order after a production crisis hits are, in practice, at the back of the queue regardless of the price they’re willing to pay. For the categories identified in the table above — HBM, conventional DRAM, high-cap MLCCs, SiC power devices — moving to forecast-based procurement with 6–12 months of safety stock for strategic parts is a substantive change in approach, not a marginal one. Cost control management tools that help balance inventory investment against shortage risk are directly relevant here.

Diversify Sourcing Channels Rather Than Depending on a Single Route

Sourcing runs the length of a product’s life — prototype builds, volume production, and the years after a part is withdrawn while a fielded design still has to be supported — and in an allocation-driven market no single route has priority stock for a buyer who was not already on a forecast. A sourcing network broad enough to reach verified inventory across multiple geographic markets and supply channels is what turns an allocation event into a sourcing exercise rather than a line stop. What decides whether that stock is safe to buy is how the supplier manages quality, and the questions worth asking are the same whoever is quoting: which certifications it holds (ISO 9001 and AS9120 are the meaningful ones for electronic components), whether it is an ERAI member, what its counterfeit-mitigation and inspection process is, and what its date-code and return policies say. This is especially relevant for the Nexperia-adjacent legacy discretes where OEM requalification fallout is still working through the system.

Track EOL and Lifecycle Status Continuously

End-of-life risk is a compounding factor in a shortage environment. A part that’s already constrained because of an allocation event is significantly worse if it’s also approaching the manufacturer’s last-time-buy window. Lifecycle monitoring — tracking PCN (product change notice) and EOL announcements from manufacturers on a continuous basis rather than only at design time — is standard practice at companies that have been through previous shortage cycles. It’s the difference between placing a last-time-buy order with adequate runway and discovering the window closed six months ago. Hard-to-find and EOL components are a specific focus area for distributors who specialize in lifecycle management, and the combination of a shortage event with an EOL trigger is exactly the scenario where having that relationship established in advance matters most.

Actively Manage Excess Inventory

The flip side of shortage-driven over-ordering is that it creates excess inventory risk when markets normalize — or when a design revision makes a previously purchased part redundant. Excess inventory is a real cost that procurement teams often underweight in a shortage panic. Excess inventory solutions — channels for offloading surplus stock at reasonable value — are a useful complement to any strategy that involves building safety stock, because they provide a path to manage the inventory if conditions change. In a market this volatile, being able to move excess positions efficiently is worth planning for from the start, not after you have a warehouse shelf full of DDR4 modules you don’t need anymore.

5. Outlook: When Does Relief Arrive?

Relief timelines vary by category, and for the most constrained areas, the honest answer is that meaningful relief is not coming before 2027 at the earliest, and in some cases considerably later.

Memory (HBM and conventional DRAM/NAND) is the longest-dated problem. New fab capacity from Micron and SK Hynix is not expected to reach volume production before 2027–2028. Micron has announced a US manufacturing investment program on the order of $200 billion, but building cleanrooms and commissioning advanced equipment typically takes years from capital decision to production wafer. SK Hynix’s planned Indiana facility — a testing and packaging site, not a wafer fab — is targeting production toward the end of the decade. Meanwhile, AI-driven demand continues to accelerate. IDC’s framing of this as a structural reallocation rather than a cyclical shortage is the more accurate mental model for planning purposes.

Legacy discretes (Nexperia-adjacent) depend more on geopolitical resolution and requalification timelines than on industrial capacity. The late-2025 truce restored nominal shipments, but AEC-Q100/Q101 requalification for automotive-grade alternate suppliers is a months-long process regardless of motivation, and many buyers are still working through it. Full normalization of this category requires both a stable geopolitical environment and completed requalification pipelines — neither of which has a firm date.

High-capacitance MLCCs face multi-year capacity build-out timelines. MLCC manufacturing requires extremely thin, precisely layered ceramic-metal stacks, and new production lines take a year or more from capital decision to output. The passive component market is expected to remain constrained through at least mid-2027, with spot prices for AI-grade parts well above pre-tightening baselines.

Advanced SiC/GaN power devices should see gradual improvement as EV and data-center-driven capacity investments mature, but extended lead times for automotive-qualified parts are likely to persist through at least the end of 2026. The 8-inch wafer transition in SiC manufacturing is creating additional yield-related pressure that will take time to resolve regardless of investment levels.

The signals worth monitoring: quarter-over-quarter deceleration in DRAM contract price growth, confirmed volume production dates from new fab builds, and sustained stabilization — not just temporary truce — in the US-China trade relationship. None of these are near-term catalysts. The practical posture for procurement planning through 2026 and into 2027 is to assume constrained conditions continue, and to build sourcing strategy around that assumption rather than around an anticipated recovery.

The Bottom Line

The 2026 chip shortage is real, but it is not uniform. For procurement teams, the most important thing is to separate the genuinely exposed parts in a BOM from the ones that are available on normal terms, and to build a response plan around the former specifically. Panic-buying across an entire BOM is expensive and, in a selective shortage, largely unnecessary. Ignoring the constrained categories because the broader market looks healthy is where production schedules get broken.

The categories that warrant focused attention right now are: HBM and conventional DRAM (allocation-driven, no near-term relief), Nexperia-adjacent legacy discretes (geopolitically disrupted, requalification in progress), high-capacitance automotive and AI-grade MLCCs (structural supply constraint through at least mid-2027), and advanced SiC/GaN power devices (lead times quoted in quarters). Everything else requires normal procurement discipline — lifecycle monitoring, occasional second-source qualification, and sensible safety stock — but is not in acute shortage.

For the constrained categories, the most effective response combines three things: early visibility into which BOM lines are exposed, qualified alternates identified before the shortage hits rather than after, and a sourcing network broad enough to access inventory across multiple geographic markets and distribution channels when a primary source goes on allocation.

Frequently Asked Questions

Is there still a chip shortage in 2026?

Yes, but it is selective rather than universal. The acute pressure in 2026 sits in HBM and conventional DRAM, automotive-grade legacy discretes affected by the Nexperia disruption, high-capacitance MLCCs, and advanced SiC and GaN power devices. Standard logic, mature analog, sensors, connectors and most microcontrollers are available on close to normal terms, so exposure depends on what is in your BOM.

Why is DRAM so hard to buy right now?

Because the three companies that make virtually all of the world’s DRAM have moved wafer capacity toward HBM and server-grade DDR5 for AI data centers, and a gigabyte of HBM consumes several times the wafer area of standard DRAM. That leaves less capacity for conventional modules, pushes mainstream lines onto allocation, and shortens the period for which a quote is valid. New fab capacity is not expected to reach volume before 2027–2028.

What happened with Nexperia and why does it matter for my BOM?

In late 2025 the Dutch government took control of Nexperia and China responded by blocking exports from Nexperia’s Chinese packaging and test operations, cutting off a major supplier of automotive-grade transistors, diodes and MOSFETs. A late-2025 trade truce restored nominal shipments, but automotive parts need AEC-Q100/Q101 requalification before an alternate supplier can be used, so many buyers were still working through second-source qualification well into 2026.

Should I switch from just-in-time ordering during a shortage?

For the constrained categories, yes. Suppliers on allocation prioritize customers who share a rolling demand forecast, and a spot order placed after a crisis goes to the back of the queue regardless of price. Forecast-based ordering with a safety-stock position for strategic parts is the practical response — paired with a plan for disposing of excess stock if conditions change, because over-ordering across a whole BOM is expensive and usually unnecessary.

How do I check whether a supplier’s stock of a constrained part is safe to buy?

Ask the same questions of anyone quoting: which quality certifications it holds (ISO 9001 and AS9120 are the meaningful ones for electronic components), whether it is an ERAI member, what its counterfeit-mitigation and incoming-inspection process is and how that is matched to the risk of the material, and what its traceability, date-code and return policies say. For automotive or safety-critical parts, specify the inspection level you need on the purchase order.

© 2026 Win Source Electronics. All rights reserved. This content is protected by copyright and may not be reproduced, distributed, transmitted, cached, or otherwise used without the prior written permission of Win Source Electronics.