The term is used loosely enough to mean almost nothing. Here is the concrete engineering definition, what genuinely changes, and how to tell a real SDV claim from marketing.
Beginner7 minSDV · Fundamentals
Almost every car company now says they are building software-defined vehicles.
The phrase appears in press releases, job adverts and architecture documents,
and it is used so loosely that it has nearly stopped meaning anything.
Note what is not in that definition. Nothing about touchscreens, connectivity,
electric drivetrains or artificial intelligence. Those often accompany SDV, and
none of them is what the term means.
To see what changed, look at how it worked for thirty years.
A car maker decides it wants heated seats. An engineer writes a specification. A
supplier builds a small computer — an — that does exactly that: reads
a switch, reads a temperature sensor, drives a heating element.
That ECU is delivered, tested, and bolted into the car. Its software is burned in
and essentially never changes. Multiply by a hundred functions and you get a
vehicle with a hundred small computers, each doing one thing, wired together.
How vehicle electronics got here — Left to right: one computer per function, then grouped by function, then grouped by physical location, then consolidated into central compute with simple zone controllers doing only the wiring.
A hundred small controllers becomes a handful of powerful ones, with simple
zone controllers near the wiring doing little more than connecting things up.
This is the change that makes everything else possible, and it is a hardware
change — which is why SDV cannot be retrofitted to an existing platform.
Software stops being finished at start of production. Updates arrive over the
air, throughout the vehicle's life, which means the development process has to
keep running for fifteen years rather than stopping at launch.
The engineering is harder, the platforms cost more, and the organisational change
is painful. Three reasons it is happening anyway:
Complexity became unmanageable. A hundred ECUs from thirty suppliers, each
with its own release cycle, is not integrable at the rate customers now expect
features.
Customers expect improvement. A phone gets better after purchase. A car that
does not now feels dated within two years.
Revenue after the sale. Features, subscriptions and services sold to a vehicle
already on the road — a genuinely new business model, and the one most often
cited internally.
The next topic goes into these forces properly, because they explain most of the
decisions you will otherwise find arbitrary.
