You have been following technological developments for many years. Which milestones were crucial for you on the way to the digital society?
Prof. Trinitis: One milestone was the World Wide Web with the principle of hypertext, which connects information via links. Tim Berners-Lee laid the foundation at CERN for what we know today as the internet. My doctoral advisor said back then: This will change humanity, just like the invention of the wheel. He was right. Over time, processors became faster and faster, and electronic circuits increasingly smaller, so that today we can process huge amounts of data – be it for scientific computing, numerical simulations, or increasingly for AI applications and machine learning.
Prof. Herkersdorf: For me, the formulation of Moore's Law in 1965 was a key moment. It states that the number of transistors on the same chip area doubles approximately every one and a half to two years. This is when the actual dynamic of digitalization began, and this development has been confirmed: Today, there are multiple processor cores, and in the case of graphics processors even several hundred cores, on a single chip. In parallel, digital communication technology developed: from Ethernet to TCP/IP to the internet. The next major chapter changing our society is machine learning and the artificial intelligence emerging from it.
Does AI have a similarly high significance as global networking through the internet?
Prof. Herkersdorf: Yes, at least. However, we are only at the beginning: the models are constantly growing and transforming. Today's Large Language Models contain billions of parameters, and this enormous amount of data places high demands on the processing of this data. Looking at our university alone, I notice that while we have ideas about how we want to use AI in teaching and research, the whole thing still needs to establish itself.
Prof. Trinitis: I see it a bit differently. Through modern technology, the computer can process ever larger amounts of data ever faster to extract information. However, this is nothing new, but merely a consequence of general technological development. Neural networks already existed in the 1980s and 90s. What we are currently witnessing is the progressive automation of all possible areas of life. From a societal perspective, this is certainly a revolution, but not technologically.
Many innovations, such as the first computer by Konrad Zuse, are initiated in Germany, whereas production or further development frequently takes place in the Asian and North American regions. What is the reason for this?
Prof. Herkersdorf: I observe that too, but I wouldn't say that this is a German or European phenomenon. Innovation alone is not enough; it also requires the right timing, the appropriate environment, and technological maturity. Take, for example, the previously mentioned TCP/IP protocols. These were introduced in the USA in the 1970s, but initially found only very limited application until the World Wide Web helped them break through almost 20 years later. Konrad Zuse is another good example: His developments were visionary, but for mass marketing, they lacked the miniaturization and cost reduction of semiconductor technology that led to the "PC for everyone".
Prof. Trinitis: In Germany, we have a tendency towards bureaucracy and a certain inertia. When it always takes so long to initiate or push things forward, it is a real obstacle. Furthermore, I consider nationalistic thinking to be extremely counterproductive. In my eyes, Europe can only survive if it acts together – both economically and politically. If each country doesn't just think about its own advantages and bureaucracy is simplified, a lot has already been done.
Ever smaller AI chips are being developed for increasingly complex computing tasks. How are Europe and Germany faring in this development? What locational advantage does Taiwan, in particular, have in international comparison?
Prof. Trinitis: There are already a few chip factories in Europe and Germany, but many of them do not belong to European companies. Intel wanted to build a factory in Magdeburg, but the project is now not coming to fruition after all. GlobalFoundries operates a production facility in Dresden. There are some European manufacturers, but technologically they do not yet belong to the world elite. However, Europe is not completely defeated. ASML in the Netherlands is the world's leading manufacturer of lithography systems (More on this on page 23). Without these, no chips could be produced. It's just that many people don't realize this.
Prof. Herkersdorf: It is often overlooked that Europe, and Germany in particular, are market leaders or even world market leaders in areas such as power electronics or automotive electronics. ST Microelectronics and Infineon are among the leading European manufacturers of microcontrollers for the automotive sector. Taiwan's success in the semiconductor sector is based on state funding, well-trained talent, and contract manufacturing. The government recognized the strategic importance of the semiconductor industry early on. Many of today's TSMC employees received their education abroad and subsequently returned to national research institutions and universities as highly qualified professionals. TSMC does not develop its own chips and is therefore not a competitor for its customers. For instance, Apple also prefers to have production done at TSMC rather than with a direct competitor like Samsung. The success of Taiwan in the semiconductor field is based on state support, well-trained talents, and the foundry model.
Do the planned AI gigafactories in Germany represent an important step toward independence from international supply chains?
Prof. Herkersdorf: Something is underway. TSMC, Infineon, and Bosch are building a plant in Dresden. Nevertheless, the overall situation will hardly change at first. They are not producing chips with the most modern structures of two or three nanometers there, but primarily 22-nanometer technologies, which are sufficient for the automotive sector. At five to seven nanometers, the mask costs alone can quickly reach several tens to 100 million euros, and the investment costs for such factories even several billion euros. These investment costs are only worthwhile at extremely high unit quantities, such as those achieved with processors in smartphones or in high-performance AI computing centers.
How is it possible to build up skills in chip design and keep or attract talent in Germany? What role can universities like TUM play in this?
Prof. Herkersdorf: The infrastructure that leading companies offer together with top universities to expand competencies in chip design at the highest international level is fundamentally very good in Germany. TUM set an example early on, for instance over 25 years ago with the first international English-language master's program for Communications and Electronics Engineering (MSCE) in Germany. Almost at the same time, TUM Asia was established in Singapore along with its Integrated Circuit Design (ICD) master's program, where our education is combined with mandatory phases in the industry. This way, we can offer the quality of TUM teaching not only in Munich or Heilbronn, but also in Southeast Asia and directly within companies. Many of the talents trained there subsequently look for the chance to continue their professional careers in Germany.
Prof. Trinitis: The crucial thing is to convey these competencies at universities. That is why I offer courses in hardware development. Furthermore, there are efforts by the former electrical engineering faculty to offer a comparable degree program in Heilbronn. Universities should increasingly focus on talent development. We cannot rely on chips coming from other countries forever and ever. To keep or attract talent, attractive offers are required. One must clarify the advantages of Europe compared to many other regions of the world – security, labor rights, freedom of research, and great opportunities for individual development. The Munich Advanced Technology Center for High-Tech Chips at TUM (More on this on page 10) is a step in the right direction. However, the focus shouldn't just be on AI chips – chip and processor development is important in general.
Where do you see the biggest stumbling blocks?
Prof. Trinitis: Very clearly in bureaucracy. One of our Japanese students, for example, waited nine months for the extension of his residence permit. If we want to retain talent, we shouldn't alienate them in this way. I once jokingly said: If it takes that long for doctoral students, their dissertation will be finished before they get their residence permit. The shortage of skilled workers is also connected to the fact that bureaucracy deters many talents. Word gets around about things like this, and it is not exactly a locational advantage.
Prof. Herkersdorf: We must take everyone with us - this concerns not only technology but also societal issues. AI creates new, highly qualified jobs; but it will also make some existing jobs in their current form obsolete or at least massively change them. Therefore, we must show future perspectives. For entire industries and even world market leaders who were successful for decades in a world without AI and digitalization, it is an enormous challenge to reposition themselves and say: "Now we're doing it differently."
What makes you optimistic?
Prof. Herkersdorf: I am very optimistic. Similar to the USA, there are now private patrons in this country who care deeply about academic education. Investments, for example by the Dieter Schwarz Foundation, complemented by research institutions like the German Research Foundation (DFG), fall on fertile ground. If the spirit of the times plays along, we can fully exploit existing ideas and potential, provided we also think in alternative directions. Using AI as an example: not always models with even more parameters, but achieving just as much with fewer parameters and new algorithms.
Prof. Trinitis: When I see the enthusiasm some students show, how quickly they learn, and how easily they are inspired, it makes me happy that I can make such a difference – even if only on a small scale. But even beyond the academic world, more and more players are waking up and realizing that we should become more independent from international supply chains. To do this, we must leave the comfortable path and develop new ideas. For us as a university, this is an advantage: we educate talents and will continue to be needed. In this respect, I am optimistic that we will continue to have plenty to do in the future.