Demand from data centers is reshaping the semiconductor market, while parts of industry are once again struggling with long lead times. New fabs in Europe promise relief, but for procurement, selecting the right components and maintaining reliable supplier relationships remain decisive.

Bottlenecks in the semiconductor market

‍

Anyone buying electronic components for machines and equipment must, in some cases, once again prepare for lead times of several months. A market overview by the distributor Globx cites 30 to more than 55 weeks for microcontrollers and 26 to 40 weeks for power and analog semiconductors. How long companies actually have to wait depends on the specific component, and a single missing microcontroller can be enough to stop the delivery of an otherwise finished system.

In many companies, this development brings back memories of the chip crisis of 2021 and 2022, when production plans had to be changed, alternative components approved under time pressure and missing quantities bought at high prices. Today, it is above all the build-out of AI data centers that is shifting demand within the semiconductor industry and thereby also influencing manufacturers' investment decisions. At the same time, geopolitical conflicts are increasing uncertainty in a production process whose individual steps are often spread across several countries.

However, the bottlenecks do not affect all semiconductors equally: while some standard components are readily available, companies have to wait several months for certain microcontrollers. For procurement, it is therefore becoming more important to track the availability of the components actually in use and to prepare alternatives before a shortage reaches their own production.

‍

The electronics industry feels the pressure first

‍

Figures from the ifo Institute for September 2026 show how differently the supply situation plays out. Overall, 14.6 percent of German industrial companies reported missing intermediate products, up from 13.3 percent in August. Among manufacturers of data processing equipment and electronic and optical products, however, the share was 37.8 percent, compared with 16.3 percent in mechanical engineering and 8.9 percent in the automotive industry. The survey also covers materials other than semiconductors, but it illustrates how strongly the supply problems are concentrated in parts of the electrical industry.

Internationally, too, companies are reporting a more difficult supply situation. In a survey by the Global Electronics Association published in August, 64 percent of electronics manufacturers reported limited availability or longer lead times for components and materials. While 44 percent had observed a deterioration in the second quarter compared with the first, only ten percent reported an improvement. Affected product groups include not only memory chips but also printed circuit board materials and passive components, so the difficulties run through several areas of electronics manufacturing.

Among participants in a Tacto survey from October 2026, 49 percent reported noticeable restrictions for individual products and 14 percent for several product lines. For three percent, production has stood still or is standing still, which underlines the economic importance of even comparatively inexpensive components. If there is no approved alternative for a microcontroller, its failure can prevent the delivery of a machine whose value exceeds the price of the component many times over.

‍

The AI business is changing manufacturers' priorities

‍

The link between the AI boom and industrial supply is less direct than the idea of a general competition for scarce chips suggests. High-performance processors for data centers, memory chips and microcontrollers for machine controls require different manufacturing processes, and their capacities cannot simply be swapped for one another. Not every additional order from a data center therefore comes directly at the expense of an industrial customer.

How strongly the AI business nevertheless influences manufacturers' decisions can be seen in the memory market. According to TrendForce, suppliers prefer to use modern manufacturing capacity for high-performance server products, while other applications have to make do with a tighter supply. For the fourth quarter of 2026, the market researchers expect contract prices for conventional DRAM to rise by ten to 15 percent compared with the previous quarter, and for NAND flash by 15 to 20 percent.

This development also affects industry, because memory is used in controllers, operator panels and embedded computer systems. Switching to other components there often requires extensive testing and approvals, and sometimes changes to hardware or software. Industrial buyers therefore cannot easily adapt their purchasing behavior to a changed supply.

On top of this come different time horizons: while semiconductor manufacturers align their portfolios with future growth, machine builders must also be able to build and service equipment whose electronics were developed years ago. If an older product series is discontinued, this can trigger a redesign at the customer that was originally planned neither in terms of time nor budget.

Precautionary double ordering makes the situation even more difficult, because manufacturers can then hardly tell what actual demand lies behind the incoming orders. As soon as supply eases, cancellations and excess inventories loom, and working them off can dampen demand for a long time. For manufacturers, this experience increases the risk of further investment, even when individual customers urgently need additional capacity.

‍

New fabs bring gradual relief

‍

Meanwhile, Europe is expanding its semiconductor manufacturing, as shown by Infineon's Smart Power Fab in Dresden, which opened in July 2026. With an investment of five billion euros, additional capacity is being created there for power semiconductors as well as analog and mixed-signal technologies. The intended applications range from vehicles and industrial systems to the power supply of AI data centers, so demand from industry and AI infrastructure partly overlaps for these technologies.

Also in Dresden, ESMC, the joint venture of TSMC, Bosch, Infineon and NXP, is building its fab. With planned investments of more than ten billion euros, the plant is designed for an annual capacity of 480,000 wafers and is intended to serve the automotive industry and other industrial customers, among others.

Before such projects improve supply, the production lines have to ramp up and the products made on them have to be qualified. In addition, extra capacity only helps where it matches the components needed, which is why a larger supply of power semiconductors does not automatically eliminate a bottleneck for a specific microcontroller.

The new plants strengthen European manufacturing, but they will not resolve international dependencies for the foreseeable future. The EU's goal of increasing its share of the global semiconductor market to 20 percent by 2030 also remains ambitious. In 2025, the European Court of Auditors concluded that the Chips Act is unlikely to be sufficient to achieve this goal.

‍

Two suppliers can depend on the same plant

‍

The Nexperia case showed how quickly political decisions can disrupt existing supply chains. At the end of September 2025, the Dutch government intervened in the company under the Goods Availability Act, giving it the power to block or reverse certain corporate decisions. When China imposed export restrictions on Nexperia products from October 4, the dispute also affected the manufacturer's international customers.

For risk assessment in procurement, it is therefore decisive not only where a company is headquartered but also where the wafers are fabricated and where the chips are assembled and tested. Even a European contractual partner can depend on a single site in Asia without this dependency being immediately apparent from the usual supplier information.

The same applies to additional sources of supply: two distributors may offer components from the same manufacturer, while different manufacturers may use the same foundry or assembly provider. If that provider fails, several supposedly independent sources are affected at once, which is why a second source should also be checked for shared production sites and supply routes.

This check must be carried out with sufficient lead time so that, in addition to the supply chain, the technical suitability can be assessed and the component approved. If the search only begins after a supply failure, an alternative often can no longer be introduced in time for ongoing production.

‍

Procurement needs a say earlier

‍

Many later supply risks are already locked in during product development, when the choice of a component determines the range of possible manufacturers and the effort of a later switch. Procurement and engineering should therefore jointly check whether the expected availability matches the planned service life of the machine or system.

Besides technical properties, the manufacturer's plans are also relevant, such as announced generation changes or long-term support for a product series. A low unit price quickly loses its significance if the electronics have to be redesigned a few years later, creating renewed development and testing effort.

An already approved second source should also be included in ongoing purchasing planning so that the supplier knows the demand and can plan capacity for it. Anyone who reserves it exclusively for emergencies must expect that the available quantities will by then already be promised to other customers.

Smaller industrial buyers in particular face the difficulty that their order volumes often carry little weight in the business of large semiconductor manufacturers. Reliable forecasts and information about product lifecycles or future projects cannot fully offset this disadvantage, but they make planning easier for the manufacturer and create a better basis for cooperation.

Direct technical exchange with the manufacturer can be combined with purchasing through a distributor. In this way, a company can obtain early information about the product strategy while also using the distributor's support for logistics, availability and inventory planning.

‍

More security has to make economic sense

‍

Additional inventory can bridge supply failures, but it ties up capital and carries the risk of no longer being needed after a drop in demand or a technical change. Long-term purchase agreements involve a similar trade-off, because they can improve access to capacity but may also oblige the company to take delivery even if its own demand falls.

Which safeguard makes sense therefore depends on the importance of the respective component for production. For a hard-to-replace microcontroller on which an entire product line depends, a larger stock can be economically justified, while for a readily available standard part with several qualified manufacturers, lower stock levels may be sufficient.

Price demands should also be assessed by product group, since rising memory prices do not automatically justify higher demands for other semiconductors. Procurement needs information about the specific market situation for this and should be able to understand which changes an announced increase is based on.

Equally important is how binding delivery commitments are, because an order placed early only creates limited security as long as price or delivery date remain open. Framework agreements and order confirmations should therefore clearly state which quantities are committed under which conditions and under what circumstances changes become possible.

‍

Limited market power makes early preparation all the more important

‍

For many of these measures, companies initially lack the necessary overview. In the Tacto survey, 19 percent of respondents named a lack of transparency about which chips are used where as the biggest obstacle to better preparedness, while 33 percent pointed to low purchase volumes and little leverage with manufacturers.

While a company's own market position can hardly be changed in the short term, companies can record their requirements and dependencies more systematically. When bills of materials, orders, inventories and manufacturer information are brought together, it becomes visible which components affect several product lines and where different plants depend on the same supplier.

This gives procurement a basis for bundling volumes, qualifying alternatives in a targeted way and building up stock where a failure would cause the greatest damage. At the same time, conversations with suppliers become more concrete when demand can be presented transparently beyond individual orders.

Industrial buyers can hardly influence the investment course of the semiconductor industry, but they can influence their own lead time. Those who identify critical components early and prepare technical alternatives can still choose between several solutions when shortages loom, before the only option left is the short-term and often expensive purchase of whatever remaining stock is available.