Autonomous driving will tip the car market. In order to still be successful in the long term, many legacy OEM and suppliers have to quickly renew their complexity.

Autonomous driving will tip the car market because the performance of artificial car brains grows with the number of autonomous vehicles. As soon as an autonomous vehicle fleet dominates, it will set the de facto standards.

The one-time expense for autonomous driving is high due to continuous and rapid innovation. The marginal costs for software updates over-the-air (OTA), on the other hand, decrease towards zero as the number of autonomous vehicles grows. Aging isolated solutions for autonomous driving with a comparatively small number of vehicles will die from incompatibility, integration effort and maintenance effort.

Today, autonomous driving is based on a specialized computer that resembles the brain of a lower vertebrate animal. This primitive artificial brain perceives the movements of the environment and the vehicle and gives them meaning in order to drive safely, cleanly, economically, and comfortably under changing conditions.

It cooperates with other cars to learn faster. To do this, it sends, among other things, 3D films of its real environment to a powerful computer for machine learning. In return, it is rewarded with changes to its software OTA. The car learns like a vertebrate animal by changing its software while it is resting. The more such cars cooperate, the faster each of them learns.

The performance of these artificial brains will increase exponentially along a roadmap. Why? The efficiency of the human brain is being better understood, machine learning is being developed by leaps and bounds, the hardware is specialized beyond Moore’s law and the performance of the system architecture, consisting of algorithms, software and hardware, is leapfrogging.

Few star engineers do this, as in the technology and aerospace industries. Unlike application engineers in today’s automotive and supplier industries, they habitually develop to physical limits. TESLA, for example, only has around 200 software engineers and 100 chip designers for the development of autonomous driving. In addition, there are 500 to 1000 developers who label the transmitted 3D films.

The business model of legacy OEM is largely based on related mechanics, mechatronics, and ECU. Suppliers develop them, produce them in large quantities, and deliver them. OEM assemble them. OEM understand neither the software development process nor the chip design process. They far underestimate how critical they are. They certainly do not attract the best system engineers, software engineers, and chip designers. Even when it comes to autonomous driving, they rely on many competing suppliers in the naive hope that everything will fit together with the SOP, just as they know it from mechanics, mechatronics and ECU. For autonomous driving, however, they have to master the system, algorithms, software, chip design and exponentially growing amounts of data along the roadmap. ECU will almost disappear. Therefore manufacturers and many suppliers have to introduce new types of complexity and reduce old ones.

With autonomous driving according to levels 4 and 5, a residual risk remains with the OEM. They have to find solutions to deal with this residual risk. For example, they can use their data to offer usage-based vehicle insurance. If they want to protect the real-driving data, they have to check whether they can integrate a car insurance and reinsure themself.

We advise you to renew your complexity for autonomous driving.