Why are chips so important, and how did it come about that we are facing a shortage of them and that they are a power issue that is driving international relations?
Tomáš Pitner is a leading Czech expert on computer systems and cybersecurity. As the head of the Centre for Education, Research and Innovation in Information and Communication Technologies, he participates in research projects in the field of semiconductor development and foreign missions, including to Taiwan, where industrial and research cooperation is arranged.
"Last time we were there, we experienced a moment that illustrates a lot. On a tour of their premises, a colleague says that we have the same equipment at home, but that his technicians say that at home in Brno, in a quiet area away from all the hustle and bustle, you can't produce better than 40-nanometer technology. Yet in Taiwan, in a seismically active area and on the second floor of a building, they showed us how their production works on 16 nanometres (shrinking the components inside computer chips is the main driver of increasing their performance and reducing power consumption, ed.). Their experience and, most importantly, their willingness to push the development envelope plays a huge role in all of this. After all, they have rebuilt places that were once just forest, mountains and sea!" Pitner reveals the background to the success of Asian manufacturers.
Where are we now as the Czech Republic or Europe with chips?
We're already panicking a bit because we've realised how little it takes to not have them and therefore not be able to function. And at the same time we know that in the future we will need them many times more, but we don't have the capacity to produce them in Europe. After all, it has only been a few months since it was not even possible to buy a car because of the lack of semiconductors. We are dependent on Asian production, and part of the problem is that Europe has set such sustainable targets that we are unable to produce chips or, for example, electric cars at competitive prices. What we are still excellent at, however, is instrumentation, that is to say, the instruments and the machines to make them. The most advanced technologies, on which, for example, Apple produces its M chips, are the machines of the Dutch company ASML. By the way, it is the Czech company Meopta that supplies ASML with optics for its high-end EUV lithography machines.
How did the current division of roles in the semiconductor industry come about historically?
Semiconductors have been known for quite a long time, but it was the invention of the transistor at the end of the 1940s that brought about the fundamental change from expensive, fragile, bulky and energy-intensive vacuum tubes. In the 1950s, transistor production spread all over the world, including Czechoslovakia. Even in Rožnov we were not that far behind. However, it got fundamentally worse in the 1970s after the construction of the first microprocessors, which were a leap forward mainly for computing and made it possible to get it on every desk. Because of export restrictions on Western technology, we started to definitely lose out. At the same time, by contrast, the Asian tigers Taiwan, Korea, Hong Kong and Singapore were investing heavily in chip manufacturing and electronics in general. An entire generation of Taiwanese experts studied in the US from the 1970s to the 1990s, and then built world-class manufacturing capabilities at home. Much of the semiconductor production was then consumed in end products in Asia as well - conglomerates such as Samsung, LG, ACER and ASUS were founded there.
And what about the resource intensity?
Manufacturing consumes large amounts of clean water and electricity, which is why Europe was happy to get rid of it, or rather not to build new capacity later on, and since the relaxed 1990s has relied on the security of international trade lasting indefinitely. But complicated permitting processes are another matter; South Korea, for example, is currently the complete opposite of the Western world in that it is trying to be as independent as possible in everything.
And when and why did we start to perceive that we, as Europe, are at a disadvantage in the chip field?
From a global point of view, the trade war between the US and the People's Republic of China over strategic technology has long been crucial. It brought an end to the almost unrestricted sharing of the most advanced technologies and at least a partial reversal of the trend of manufacturing shifting to mainland China. The situation in Europe is also indirectly affected by the fact that, in the light of the US-China rivalry, direct Taiwanese investment in mainland China has also been sharply reduced. Free resources, together with the need to expand production, have brought interest in Europe. In this, governmental or European financial support is more of a necessity than a competitive advantage, as they all provide it.
Moreover, the demand for semiconductors is still growing. Where?
Telecommunications is still a key customer, and that's an awfully broad area – from network infrastructure to transmitters to the mobile phones we carry in our pockets. It still includes computing, of course – the computers we have on our desks, but also the computing infrastructure: data centres, large storage facilities. A perennial classic, but with a noticeable evolution, is renewables, especially photovoltaics, which is explicitly based on semiconductors. And the further boost in demand for semiconductors is caused by the massive growth in electromobility. For all the sophistication and digitalisation of conventional cars, EVs are still heavily dependent on semiconductors - everything is digitally controlled.
In general, all drives today are semiconductor-based, no matter whether you need to turn a motor in a drill, an electric car or a locomotive. Many of the technologies used today did not exist 15 years ago. In addition, the energy industry will be stepping up its demands. Consider that we typically generate electricity in the north of Germany offshore, for example, but we need to get it to the Balkans. Long-distance transmission is a huge bottleneck, and across the sea in particular. This is because the current needs to be converted to direct current for underwater HVDC lines and then made alternating again by inverters. Semiconductors take care of all that.
For years there has been talk of the dominance of two computer chip manufacturers, Intel and AMD. The former in particular is now struggling, and NVIDIA is on the rise. Why is it that forces are realigning?
The biggest chip maker in general is TSMC (Taiwan Semiconductor Manufacturing Corporation). The two companies mentioned have been competing for decades, but Intel has always been the one with the bigger sales and market share. The problem with both is that they have been making them for years. They have to hold on to older lines for spare parts while wondering what will sell in 10 or 20 years. This is the arsenal that Intel is now dragged down by. NVIDIA is a completely different story. It originally was established for games and virtual reality, graphics and multimedia, and it seemed like it was going to grow in that segment. Normal CPUs have, say, 16 cores and 32 threads, whereas the graphics ones have multiples of those numbers and the cores are in the hundreds. It turns out that for certain types of computing, like when you need to pan the image so you do the same thing with many pixels, these products are more suitable than regular chips. It turns out that this is the nature of the tasks associated with AI learning, and NVIDIA has already been great at it.
What's actually the trend in chip research and development?
What's worth watching is Apple's approach. The company has made a significant breakthrough with the concept of custom A, S and M series processors that are fitted into their devices based on the performance required. Apple has correctly assessed that it pays not to be dependent on an external manufacturer (until then Intel) and, in particular, with the high-end M series that it is installing in computers and laptops, we see several trends:
Apple designed them but doesn't directly manufacture them, they are made on the most advanced 3nm technology at the moment in TSMC's factories. Higher integration means smaller circuits and lower losses, so laptops last longer without power, but the need for noisy cooling is also eliminated. In addition, maintaining multiple compatible, co-developed processor families for different sized devices makes it more efficient and reliable to write applications that will run on the devices. This level of platform unification has never been seen before, and other manufacturers can be expected to follow suit. In addition, the unified RISC-V architecture supported by the European Union and design tool vendors such as Codasip, developed in Brno, can help greatly.
And how do you feel about the often mentioned vision that we as the Czech Republic can become a significant player in this field?
It's not an exaggeration, we have a decent starting position. Production capacities will expand, there is a good tradition in education as well. Our experience in manufacturing and engineering is really unique. Massive investments at our neighbours in Germany (especially in Saxony, where TSMC is heading) will help us, and a similar network of subcontractors may emerge in the Czech Republic as we have learned in the automotive industry. But we must not lose this by the usual sluggishness, slowness and resortism. I am quite optimistic, I believe that the semiconductor sector in our country will succeed in the same way as the IT industry now or the engineering industry even earlier. After all, anyone who sees the Czech industry teetering on the edge of decline senses that an economic recovery of classic industries is unlikely without generational modernization.