Minimize Complexity How Car Manufacturers Will Win the SDV-Race

Anbieter zum Thema

congatec GmbH

Apacer Technology B.V.

Caltest Instruments GmbH

There’s no shortage of ideas for next-generation vehicles. The question is how quickly those ideas reach customers, and how they improve over time. That comes down to something too often overlooked: architecture.

Over the last decade, ADAS, electrification and the connected cabin pulled software to the center of the vehicle. Features have been racing ahead ever since. Architectures, in many cases, have not. That gap now shows up as program drag, cost growth and integration risk.

Why the “add another ECU” Model breaks under SDV Pressure

For years, adding capability meant adding another ECU, often with different silicon, software and integration models. Across a vehicle, that creates validation overhead, unpredictable updates and more interface stitching than new value. What once worked now becomes the bottleneck when customers expect frequent over-the-air improvements.

The fix is not “more boxes.” It’s decoupling software from hardware and running on a unified foundation that delivers consistent behavior across the vehicle. Just as data centers tamed runaway complexity by standardizing compute and orchestrating systems together, automotive must now shift from hardware sprawl to a unified architecture designed for predictable behavior at scale.

What a Unified Vehicle-core Architecture Looks like—and why it Works

A vehicle-core architecture coordinates how the car behaves: motion, energy and thermal management, body systems, in-vehicle networking and security. Around that core, automakers still keep tight loops at the edge and organize I/O in zones, but the system is logically unified and architecturally disciplined:

• Central domain: where real‑time, safety‑critical coordination lives and timing must be deterministic.

• Zonal layer: where functions and I/O are grouped physically, simplifying wiring and enabling orderly data and energy distribution.

• Edge nodes: where low‑latency control stays close to sensors and actuators without fragmenting overall behavior.

Domains such as infotainment and automated‑driving feature stacks evolve at “consumer speed”. They operate outside the vehicle‑core architecture and interface through well‑defined boundaries. That separation protects real‑time, safety‑critical systems while letting fast‑moving domains innovate freely.

It is also important to consider how AI fits into this architecture. AI delivers most value where data and computing are consistent. A unified vehicle-core architecture provides both, with computing placed at the edge, in zones or at the core based on latency and safety needs.

This is also what allows AI to move beyond isolated features. With consistent data, deterministic timing, and controlled deployment, AI can improve vehicle behavior at a system level—compounding benefits across the fleet over time.

Software and Middleware: the Difference between Components and a System

Middleware is what makes SDVs work (Figure 1). A modern vehicle needs a consistent software base and a deterministic middleware layer that handles safe scheduling, mixed criticality orchestration, communication discipline, isolation and well-defined service interfaces (Figure 2).

Instead of repeating the ECU by ECU integration cycle, car manufacturers are now looking at architectures where the core hardware, networking, energy management and middleware already behave as one.

Once the architecture is in place, continuous integration, testing and deployment (CI/CT/CD) accelerate progress only when a consistent execution environment ensures results remain repeatable and reliable.

Build versus Buy: where to Invest Scarce Engineering Time

There is pride in building from scratch, but there is also a cost. Recreating timing models, safety scaffolding, OTA pathways and integration layers pull attention away from work customers notice. A more pragmatic pattern is to start with a proven foundation and focus internal teams on experiences and data-driven features.

To operationalize this shift, the industry is moving toward system‑ready SDV foundations—architectural baselines that remove the need to rebuild infrastructure program by program.

For automakers, this directly lowers total cost of ownership by reducing redundant integration work, shortening validation cycles, and enabling software reuse across vehicle lines and generations. By freeing teams from foundational integration work, these approaches let engineering effort concentrate on differentiation customers actually experience: features, services, and intelligence delivered over the vehicle’s lifetime. That’s important because customers do not ask for an SDV. They ask for range, safety, comfort, personalization and a car that is better next year than it is today.

The Bottom Line

The industry has reached the point where adding features without architectural discipline only increases cost and risk. The winners will build on a deterministic vehicle-core foundation, standardize middleware and run a delivery model – spanning deployment, validation, and updates — that allows AI-driven improvements to be introduced safely and continuously over the vehicle’s lifetime.

That is how software, hardware and operations align so intelligence can spread across the vehicle predictably and at scale. Build on that foundation, then the experience – services, personalization, efficiency – stops being aspiration and starts being standard practice.

 (se)

* Jan-Philipp Gehrmann, VP of Marketing & Strategy, Automotive at NXP

(ID:51011645)