SDV is expensive, disruptive and organisationally painful. Four pressures explain why car makers are doing it anyway — and knowing them makes most architecture decisions predictable.
Beginner5 minSDV · Industry · Strategy
Rewriting a vehicle's entire electrical architecture is an enormous, risky,
expensive undertaking. Car makers are conservative organisations. So why is
essentially the whole industry doing it at once?
Four pressures. Once you know them, most of the technical decisions you will
meet stop looking arbitrary.
Traditionally a car generates revenue once, at the point of sale, plus servicing.
Software changes that. A feature can be sold to a vehicle already on the road.
Subscriptions, one-off unlocks, upgrades — all delivered without anybody visiting
a workshop.
Pressure 4 — Regulation now requires update capability#
This one is less discussed and increasingly decisive.
Cybersecurity regulation in several major markets now requires
manufacturers to monitor for vulnerabilities and respond throughout a vehicle's
service life — which can be fifteen years.
Pressures explain the motive. Three things made it possible now rather than a
decade ago:
Automotive-grade processors got powerful enough. Running several operating
systems and dozens of functions on one chip, within an automotive thermal and
temperature envelope, is recent.
Automotive Ethernet matured. Moving from broadcast at kilobits to
switched Ethernet at gigabits — with providing timing guarantees — is
what allows a central computer to talk to everything fast enough.
Hypervisors became automotive-viable. Running a safety-rated system and a
general-purpose one on the same chip, with provable isolation, is what makes
consolidation acceptable to a safety engineer.
