Process technologies are increasingly tailored to specific uses

In spring 2026, Patrick Vandenameele will succeed Luc Van den hove as CEO of imec. In an interview with Markt & Technik, he shares his vision of the future of imec and the semiconductor industry in Europe, emphasizing the need for Europe to catch up in the field of AI.

Markt & Technik: In your opinion, was the merger with iMinds particularly important for imec? Why?

Patrick Vandenameele: Absolutely. The merger with iMinds was a decisive step at the time, and it remains so today. With the advent of artificial intelligence, there is an increasing need to integrate hardware and applications. The demands placed on hardware by applications are increasing and becoming more diverse. In the past, the industry would have been satisfied with improvements to NMOS and PMOS transistors. Nowadays, next to these improvement, increasingly it is novel system architectures — i.e. how software, memory, electrical and optical interfaces interact — that matter.

The answers to these questions depend on the algorithms used. This is precisely the area in which iMinds specializes. Thanks to the merger, imec is now able to develop and optimize hardware and software together. This was a major win for both parties: iMinds gained direct access to industry and became better known, while imec built a bridge between research and industry.

Two years ago, we emphasized the importance of this connection again and created a separate 'AI and Algorithms' department, comprising around 1,000 employees who support all our research and application fields.

Europe is currently facing an uphill battle when it comes to AI. Most of the hardware comes from the US, as does the hype surrounding OpenAI. So where does that leave Europe?

Europe remains strong in terms of research. However, when it comes to the hardware side of AI, i.e. the actual computing basis, Europe largely vacated the field years ago for a variety of reasons, including a lack of political support for scaling semiconductor technologies.

It is all the more impressive, therefore, that companies such as ASML are among the world's leading equipment suppliers despite these conditions. Such companies are key to keeping up in this field. Germany, Austria and the Netherlands also have successful manufacturers of equipment and materials that are at the forefront of the industry.

In recent years, we have significantly intensified our cooperation with these companies, supported by the European Chips Act. It is crucial that we consistently build on our strengths.

Ideally, we would like to see modern semiconductor manufacturing in Europe, such as a fab with the most advanced technologies. The most realistic way to achieve this would be to convince an existing foundry to invest here. However, demand for these manufacturing processes is currently low in Europe.

It's a recurring problem...

Yes, but things are changing, even for us. IC-Link by imec was originally intended to support universities, but it has evolved into a central platform for building European design expertise, lowering the threshold for SME’s and larger companies. Today, 3 nm and 5 nm designs are being created there, and we are preparing for 2 nm processes.

This is crucial for companies such as Axelera AI and Openchip, as it gives them access to state-of-the-art resources. While planning a 3 nm design may sound simple, when you need 50 highly specialized backend designers to execute it, it quickly becomes challenging. This is precisely the area in which we can help.

Our goal is clear: to bring this design expertise back to Europe. One area we have chosen to focus on is the automotive industry, which is highly relevant to Europe. The industry itself has asked us to help build an ecosystem for chiplets. This is a great opportunity to bring the ‘intelligence’ of the car – i.e. AI-based systems – back to Europe. Almost all ADAS algorithms today are AI-based.

Another area of focus is PDKs (process design kits). Until now, researchers have found it difficult to access the latest process technologies. This is why we have developed virtual pathfinding PDKs, which allow experimentation of real designs based on technology that we expect to be available in the future. This will enable researchers and start-ups to implement creative ideas and perhaps even resulting in the foundation of new companies.

In the past, universities used to complain that it was too expensive to turn design ideas into real chips. Is that still the case?

No, it isn't. We offer an MPW service via IC-Link by imec. We are also collaborating with several European start-ups on these cutting-edge technologies.

The European Union supports this through its Design Enablement Teams (DETs), which promote start-ups by financing design and EDA tools, for example, or manufacturing. IC-Link is a DET that focuses on the smallest process nodes. These programs have significantly lowered the barriers to entry for young companies.

Imec is renowned for its close collaboration with the semiconductor, materials, and equipment industries. However, you also want to work more closely with system companies and start-ups, and imec.istart is already geared towards start-ups. Where do you see room for improvement?

imec.istart is our accelerator program, similar to Y Combinator in the USA, albeit smaller. While it is very successful, it primarily focuses mostly on digital start-ups rather than deep tech.

This is why, around nine years ago, we founded our own deep tech fund, imec.xpand. Together with this independent fund, we run an incubation initiative, that focuses on deep tech spin-offs based on imec technologies.

Over the years, we have noticed an increasing number of very young start-ups, not spun off from imec, seeking contact with us at an early stage. In the past, we would not be able to collaborate with them due to their limited financial capacity.  However, we now recognise that innovation primarily occurs in start-ups. It would be a mistake not to support them.

That's why we now also invest in external start-ups, acting as both a financial and a strategic investor by offering our services on start-up-friendly terms.

Could you give an example?

One well-known example is Celestial AI. The company is now worth several billion dollars and specializes in photonic interconnects for AI. The component that forms its unique selling point was developed as part of our photonics interconnect program at imec.

We supported Celestial AI in developing a prototype, guided them through the investment process, and are currently preparing to transfer the technology to our foundry partners for volume production.

What advantages do you think this has for imec?

Quite simply, we learn. A successful start-up needs three things: a relevant problem to solve, the right people, and capital. That's not trivial. Having build four start-ups myself, I know how many skills it takes to be successful.

Imec brings technological expertise, but we don't always understand the specific challenges of each market. Not all of our best researchers automatically become founders, either. This is why we provide support through our deep tech fund, imec.xpand, which currently manages over 400 million euros in capital and is one of the largest in Europe.

However, this is still small in global terms. This is why we also collaborate with regions that have well-established capital structures. This international cooperation works both ways: some start-ups establish locations with us to further develop technologies together. At the same time, experienced imec engineers move to these start-ups and later return with new insights, such as how to establish a genuine start-up culture. Our major industrial partners also benefit as new opportunities for collaboration and acquisition arise.

We intend to promote this dynamic even more strongly in future.

You also mentioned partnerships with systems companies. Are you referring to tech giants such as Google and AWS, or to industries such as automotive?

Both. The requirements of the various fields of application are so diverse today that they directly influence technology development. In the past, a single technology concept could serve many markets. Today, that is no longer possible. Even large foundries such as TSMC, Intel and Samsung are developing several process variants in parallel, and the diversity in the memory sector is even greater.

If we don't work closely with system houses, we won't know what hurdles they are facing. But that is precisely our job: to remove these hurdles. Many system providers focus on their product, such as AI, while taking technology for granted. We, on the other hand, continue to develop the fundamentals, such as new transistor types, materials, thermal concepts and integrated energy management in packaging technologies.

This is one of imec's key advantages: we think ahead in terms of technology, and in collaboration with those who will use it later.

For years, discussions have been taking place about topics such as Power Wall and STCO, i.e. the integration of systems and technology. Is this really new to systems houses?

No, of course they are familiar with it. However, consider the reality: in these companies, 95 per cent of resources go into the next product, four per cent into the generation after that, and less than one per cent into long-term, disruptive technologies.

Developments that will only be market-ready in five or ten years' time are simply too far away for them. This is exactly where imec comes in. We work on long-term issues to ensure that our partners, including foundries and system houses, are aware of the technological developments that lie ahead.

Would these companies even be willing to collaborate with competitors?

That is indeed a challenge. This is why we have restructured imec into internal sectors. This gives our research teams more freedom to tailor programmes to the specific requirements of individual partners.

For example, in the automotive industry, we are working on projects that have already reached a high level of technological maturity. Our focus is on chiplet systems scheduled for series production in 2030, which is a short time horizon for us but a suitable pace for the industry.

Several original equipment manufacturers (OEMs), intellectual property (IP) providers, electronic design automation (EDA) manufacturers, foundries and packaging specialists are jointly developing functional 2.5D chiplet systems. This may not be as spectacular as quantum or AI research, but it is crucial because reliability is paramount in this area, and this can only be achieved if all players along the value chain work closely together.

However, it took the automotive industry a long time to start working in this way. Are cloud and computing companies ready for this?

Yes. These companies are system providers, but they are also increasingly developing their own chips. This makes them ideal partners for us. This industry can only solve its challenges by working together.

Our role is to create structure and coordination — in other words, to link research, development and production together to create marketable systems. This significantly shortens development cycles and enables projects to run in parallel, even when they are more complex.

This requires us to have a much better understanding of the applications. These types of partners will be crucial for imec to remain relevant in the future.

As far as design is concerned, this is not a pre-competitive area. Here, we work in bilateral collaborations, i.e. targeted alliances between individual companies.

There is currently a lot of talk about resilience in Europe, even beyond the world of semiconductors. In your opinion, in which areas should Europe strive for greater independence?

Complete independence in the semiconductor industry is almost impossible, as it is a global and highly interconnected industry. Even global chip acts do little to change this. The same applies to cloud sovereignty.

However, Europe needs its own 'trump cards' to remain capable of action. This includes building on existing strengths, such as equipment and materials.

A hypothetical competitor to ASML would jeopardise our resilience. That's why we need to build on our existing strengths.

At the same time, we should have the courage to dream again of advanced semiconductor manufacturing in Europe. When I was a student, I completed an internship at STMicroelectronics, which was the global leader in scaling at the time. It would be appropriate to revive such ambitions.

In addition to semiconductors and the cloud, artificial intelligence is, of course, also central. There is a lot of hype, yes, but no bubble. AI is here to stay. If Europe wants to play a role in this field, it must invest in AI hardware. This is an area in which we at imec can contribute significantly: we understand which technologies are scalable and how to develop them.

We can help implement such projects through collaborations with large foundries or our own European production capacities.

However, this will not come cheaply. Anyone who wants to regain a competitive advantage will need to invest tens of billion euros, just to get started.

The debate about cutting-edge manufacturing in Europe is nothing new. Do you still see any realistic opportunities for Europe?

Yes, but only if we start thinking consistently in the long term. Projects like this usually fail due to a lack of political determination. If Europe can demonstrate the necessary stamina, however, we can succeed.

At imec, we work with cutting-edge technologies every day — it's what we do. However, it is questionable whether the average European sees the same value in this. Nevertheless, it would be necessary because such a project would require substantial public investment.

Where do you see imec in four to five years?

Semiconductor technology scaling is more relevant today than ever before, especially with AI. We are developing more compact, energy-efficient technologies. Initially, many thought AI was a 'second internet' – a topic for software companies. But today, it is clear that the real challenges lie in hardware. Therefore, we will continue to expand in this area. The response has been tremendous: our recent Partner Technical Week (PTW) had over 830 participants, setting a new record.

At the same time, the importance of IC Link continues to grow. It connects technology development and application, and is central to developing European design expertise. Having come from the design sector myself, I am convinced that imec's influence will be more strongly felt in the future through real-world applications.

In classic research collaborations, such as those with TSMC or Intel, we lay the groundwork that is then further developed elsewhere. However, in areas such as health, automotive and robotics, we need to achieve a higher level of maturity ourselves and deliver marketable prototypes.

For instance, imec technology is now widely used in genome research. However, publishing scientific articles is not enough; we also need to deliver functional chips that can analyse blood samples.

All projects requiring such a degree of maturity are grouped together under IC-Link by imec. While IC-Link originally stood for ASIC design services, today it also encompasses photonics, 3D integration, and life science platforms – in other words, all areas where research directly translates into applications.

Some time ago, imec developed a cell sorter...

It was an exciting project for cell therapy, and a good example of how difficult it is to turn technology into a practical application. While it was very promising from a scientific perspective, it has not yet made it to market.

This demonstrates that, without sufficient resources or structures, such as start-ups, it is almost impossible to bring an idea to market maturity. We still consider the technology to be excellent, but without implementation, it is of little use.

Is this a specific problem in the healthcare industry?

No, it affects all sectors. Even in the automotive industry, for example, the willingness to invest in research is limited. Projects often have to be very advanced before they are taken on.

The life sciences industry is a particularly interesting field. Open innovation is difficult to implement there because the business model is based on patent protection for individual molecules. Nevertheless, some exciting approaches are emerging.

One example is our research into microphysiological systems (MPS). Rather than using animal testing, we are developing synthetic models. We generate induced pluripotent stem cells from a skin sample and use these to create a digital replica of the human body on a chip. This enables us to test new active substances quickly and efficiently without the need for animal testing.

We collaborate closely with partners such as Merck, who have the resources necessary to translate such technologies into real products. This is an excellent example of how science and industry can collaborate to drive innovation forward.