Semiconductor ecosystem: Study describes EU problem areas

Contents

After the Corona chip crisis, the Suez Canal crisis around the Ever Given, the AI demand crisis, and most recently the raw materials crisis caused by the Iran war, one thing now seems to be generally accepted: “Microelectronics is also geopolitics,” said Gitta Connemann, Parliamentary State Secretary in the BMWE and CDU politician at the ZVEI Summit in Berlin. It took place on Wednesday and Thursday. And this necessitates a strategic positioning, as the EU has been attempting for years. In parts of the semiconductor value chain, European independence has already been achieved, says Connemann. But now it's about closing the gaps.

However, the most important question for a business case remains demand, says Connemann – and then it's about whether European orders can be placed. “Without technological sovereignty, there is no digital sovereignty,” says Tanjeff Schadt from the consulting firm Strategy&. But without microelectronics, there is no technological sovereignty. The strategy consultancy, commissioned jointly by ZVEI, the Dutch association of the technology industry FME, and the German and Dutch ministries of economics, attempted to determine what semiconductor industry policy could be sensible in the EU. For this purpose, semiconductor demand for use in the EU, both as an internal market and for export, was analyzed.

Industrial demand rising faster than consumer demand

By 2040, demand in the core will double, according to the study. Demand from the EU industry is rising even faster than that from consumers, with certain types such as GPUs and CPUs increasing to seven times the level of 2025, and memory to three times. Datacenters are the largest growth area, with demand increasing elevenfold to the level of the automotive industry, consultants predict. However, the depiction of geopolitics is one of the factors that, according to the study, will also shape demand in the coming decades. This is because, in particular, use in critical infrastructures and security-relevant application areas, such as military applications, is a major driver of demand, requiring corresponding capacities.

However, as long as supply chains function, the cost factor currently works against Europe. Production in Europe has a cost disadvantage of 15-30 percent, consultant Schadt presented at the ZVEI Summit in Berlin on stage. For packaging, the EU is 20-30 percent pricier, partly due to labor and construction costs. This is not too far off to close these gaps, says Tanjeff Schadt. Political stability is a factor that investors also view positively.

The study is explicitly “not a call for a planned economy,” says the Strategy& consultant. Specifically, his research recommends three different stages: Maintain, Strengthen and create new capabilities, or enter into new partnerships.

Strong differences between segments

The authors apparently assume that Europe is comparatively well-positioned in semiconductor production equipment – from ASML to Zeiss and other specialists. However, in design capabilities, the authors see further demand for power and communication chips, sensors, and microcontroller units (MCUs), but the greatest need is in CPU, GPU, and SoC design capabilities. In concrete production, sizes below 7 nanometers, the category of 8-16 nm structures, and 21-46 nm structures are particularly urgent, according to the study's findings. However, more needs to be done beyond that – up to the 350 nm category.

In packaging, the authors see an urgent need for action, especially with new methods such as stacking chiplets in a confined space. For all fields they consider strategically particularly urgent, this means “Build or Partner,” i.e., do it yourself or find a reliable partner.

With this, the authors mark fields in which EU companies have less than 8 percent market share and at the same time a significant increase in demand is expected. Photonic, neuromorphic, quantum, and new memory technology chips are also in this category, though not specified further.

On the other hand, the authors consider the necessary printed circuit boards, onto which chips are mounted for end products, separately. In the past, this technology, which was not considered overly complex, was often overlooked when it came to the overall manufacturing process of industrially recyclable chip production. By 2040, automotive will continue to account for half of PCB demand. However, this lion's share is contrasted by a growing share of specialized PCBs, for example for AI servers or military use, where resilience is primarily concerned. Here, there is a massively increasing demand for capabilities that are driven not primarily economically, but strategically.

Basis for new edition of the EU Chips Act

The Dutch-German study comes not by chance shortly before the so-called “Tech Sovereignty Package,” in which the European Commission will present its plans in the week after next on how the EU is to become more independent from third parties. This also concerns the future strategy for promoting and attracting development and manufacturing capacities – for example, in the successor to the Chips Act, but also regarding AI infrastructures.

The fact that the EU has only achieved partial success with the first iteration, which was intended to lead to the establishment of Intel in Magdeburg (chip manufacturing) and Wroclaw (packaging), among other things, is to be addressed as a learning experience. Major steps towards “fabless” are to be taken. This concerns independence from specific chip companies towards contract manufacturing with corresponding service providers. However, this can hardly provide any security against supply chain problems with raw materials such as helium recently.

(dmk)