My first project as a professor with a German car manufacturer was in 2013, and our industry partner kicked off the first project meeting with the words:
Ich bin ein Blechbieger und verstehe von Software nicht so viel.
(I bend sheet metal and don’t know much about software.)
I think he already was behind the times and today, German car manufacturers have a good understanding and corresponding capabilities to build software for the car, for example, for the rather comprehensive infotainment stack (the center console).
The problem? That’s not where the action is. New cars today already bundle the physical device with services in a data center.
Today, the data center is riding shotgun, in the future, it will be in the driver’s seat.
Developing safety critical software for embedded devices (today’s and past cars) is wholly different from developing data center software. Amazon reportedly deploys software changes into production several thousand times a day. This closes the feedback loop on how well things work , leading to an innovation speed impossible to achieve if physical devices weigh you down .
From my projects, it is clear that OEMs are missing the capability to build data center a.k.a. cloud computing software. Degree programs, including those at my university, don’t teach this either.
The car itself will remain the place for functional safety, but for most, this is already a commodity and not a reason to buy. The competitively differentiating value will be streamed from the cloud , and OEMs need to learn how to build those services. Otherwise, Silicon Valley once more will eat some industry’s lunch.