Why Semiconductors are essential for the software-defined vehicle

In his opening speech, Congress Chairman Ricky Hudi described the shortage of chips and transistors that has plagued the industry for the past two years as one of the “key trends” in the automotive industry. Semiconductors will become even more important for the development of cars in the future. Their importance is already evident from a simple and obvious context: If the car of the future is software-defined,then semiconductor components will be the foundation for this concept. After all, every piece of software needs a piece of hardware on which to run. In the past, this was – from the perspective of a software-defined car – a comparatively simple microcontroller. Today, it requires a complex system-on-chip (SoC), a semiconductor device comprising multiple processor cores and customized task-specific hardware accelerators. In the last four years alone, the performance of these centrally important components has increased by a factor of 50, as Karsten Michels, Head of Productline HPC at Continental Automotive, explained. In the same period, the memory configuration of high-performance computers in cars increased 16-fold. And the rapidly growing transistor count of these computers drove up energy consumption by a factor of 25 – despite all the efforts of semiconductor technologists to develop energy-efficient chips; today's domain computers can easily consume up to 150 watts.

However, decisive developments are by no means taking place only in SoCs for high-performance computers (HPCs). In line with the general electronics trend toward shifting intelligence to the edge, mixed-signal components are facing significantly increased demands in the context of the digitization of the car: They are increasingly responsible for functional safety as well as for cyber security – and they must be adaptable should the need for changes or expansions arise during the life of the vehicle. This was pointed out by Jan Dienstuhl, CSO of the semiconductor manufacturer Elmos, in his presentation “Enabling the Edge: How do IC Solutions become safe, adaptable and secure”. In the sensor and actuator-related application areas, Dienstuhl sees the use of mixed-signal semiconductors growing faster over the coming years than in most other areas of the vehicle, namely by 30 percent. By comparison, the Elmos CSO (who is also responsible for development) expects the use of standalone microcontrollers to increase by only 6 percent (by number of units), the use of power components by 29 percent. Only in the case of SoCs for domain and central computers Dienstuhl predicts even higher growth, namely 150 percent.

Dienstuhl sees driving factors for the use of mixed-signal chips in power management, sensor-related data preprocessing, driver circuits for motors, displays and LEDs, as well as in system base chips and bus transceivers. In order to accelerate development cycles on the one hand, and on the other hand to be able to make functional extensions and corrections to the finished and delivered car, these mixed-signal components are increasingly becoming smart as they contain on-chip processor logic. As a result, their software inventory is also growing. The benefit: Where previously the laborious integration of circuits at the hardware level demanded the sweat of the developer, in the future the software contained in the components will make it possible to work much more flexibly.

Panel discussion on semiconductors in the automobile

The panel discussion with prominent industry experts, moderated by Alfred Vollmer, editor-in-chief at the German trade publication AUTOMOBIL-ELEKTRONIK, also addressed the role of chips for cars – and the difficulties of stocking up on sufficient quantities. The topic is now making waves far beyond the trade media. “Never before in my professional life I have seen so many media also outside the trade press dealing with this topic; now even politics have become active in this matter. Wasn't this long overdue and why?” asked Vollmer the assembled experts. “Indeed it was,” affirmed panel member Magnus Östberg, Chief Software Officer of Mercedes Benz. “It shows how strong this industry of semiconductors is. And now governments have started to recognize the situation,” Östberg added.

Does it make sense and is it feasible for several customers to join forces and buy their chips in joint bulk orders from the manufacturers? There was clear skepticism among the panelists on this question – from the issue of transparency to antitrust concerns. Only Bosch had fewer problems with this; the company manufactures a large part of its semiconductors itself. “We are of course bundling our capacities within the company,” said Jens Fabrowsky, member of the automotive division board at Bosch. On the technical level, standards and interfaces are important, as well as on the organizational level, the early indication of corresponding needs.

Lars Reger, Executive Vice President of NXP, also addressed the supply chain issue in the discussion, pointing out that the “local singularities” of this supply chain – such as the concentration of large parts of IC production in Taiwan – pose a problem. However, since the Russia started the war in Ukraine, he said, public attention has increased significantly: “A lot of people have woken up.”

Especially at the beginning of the semiconductor crisis, the competition of the car industry for the manufacturing capacities of the chip industry had become obvious. So there is still the question of how to ensure the availability of the coveted electronic components at all times. The answer came from Lars Reger, and it may well be interpreted as a reminder to the customers of OEMs and tier-one suppliers: Careful planning is the solution, said the NXP manager, coupled with the hint that even in the 15 years ahead the absolute lion's share of chips will still be manufactured in semiconductor technologies “north of 20 nanometres”. In other words, in geometries of more than 20 nm – which puts the alleged competition between consumer electronics and the automotive value chain, often cited as an argument for the supply problems, into perspective.

Are automotive OEMs now developing their own chips?

Another question that continues to plague the industry: What is the truth behind the regular reports/rumors that this or that carmaker is planning to build its own ICs? A difficult question, because no carmaker likes to be peeked into so deeply. Mercedes manager Östberg, being the only representative of the OEMs in this discussion, elegantly maneuvered around a too detailed answer: It is always about the right balance between the features, for example between (electrical/electronical) power and computing performance. In any case, Östberg said, his company would not become a silicon producer. “We are exploring how deep we want to go into the topic.” Jens Fabrowsky chimed in here. “Semiconductor manufacturing always has something to do with economies of scale. The question is: Will an OEM ever be able to afford a huge, expensive device exclusively for one brand?” Lars Reger added that not even Tesla develops all its own chips, but only an AI accelerator.: “In fact, Tesla buys 99 per cent of its chips.”