"The way most vehicles are defined today, their software and electronics layout, is you have hundreds of little, tiny computers—the industry term is ECUs, electronic control units—that have a little island of software that performs some domain-based function. … To do something like an over-the-air update requires all this coordination amongst third parties, and you have all this unused capacity, because you have all these little computers, none of which is fully being utilized."
RJ Scaringe, Rivian CEO
The automotive industry is undergoing a fundamental shift where vehicles are evolving from hardware-defined machines into software-driven platforms. In this model, intelligence, functionality, and performance are increasingly determined by software rather than merely hardware innovations.
This shift toward software-defined vehicles (SDVs) represents a structural change in how cars are designed, built, and improved over time. This blog discusses how the core of this transformation is made possible by three critical enablers, which are scalable electrical/electronic (E/E) architecture, virtualised ECUs, and innovations in semiconductor chip design. If you want to read about SDVs in more detail, check out our report, Software-defined vehicles: software takes a front seat in the automotive industry.
For a long time, vehicles have relied on distributed electronic systems, leading to growing complexity with up to 150 ECUs and extensive wiring in vehicles. Software-defined vehicles (SDVs) require a more scalable and flexible architecture, making zonal control units (ZCUs) increasingly important. Essentially, ZCUs consolidate multiple ECUs into one or a few high-performance computing units, thereby reducing software and wiring complexities. A notable example is Rivian, which has adopted a zonal architecture instead of the traditional domain-based approach.
ZCUs reduce the number of single-function ECUs, simplify wiring, reduce vehicle weight, and help shorten development timelines. By consolidating multiple vehicle functions into fewer high-performance computing units, they support centralised computing, improve energy efficiency, and reduce overall system costs. However, OEMs must ensure seamless integration of safety-critical functions through unified communication protocols.
A range of automotive OEMs and vendors are working to enable ZCU deployments or have deployed ZCUs. Infineon, Ecotron and BMW are three leading such companies and are profiled below.
Infineon is enabling the shift to zonal vehicle architecture through its semiconductor solutions for zone controller units. These ZCUs aggregate communication, power distribution, and local control functions within a vehicle zone. By consolidating multiple legacy ECU functions into fewer controllers, the company helps OEMs and tier-1 suppliers to reduce E/E complexity, wiring harness weight, and system costs.
The company uses ZCUs that consolidate body and chassis functions into structured zonal domains for vehicle platforms. The architecture reduces wiring complexity, integrates multiple communication protocols, and provides a scalable foundation for software-defined vehicles.
BMW is developing the Neue Klass of electric vehicles, incorporating ZCUs and with an architecture that allows 600 meters less wiring and 30% reduced wiring weight.
In addition to architectural changes, automakers are ramping up efforts to deploy virtualised ECUs that behave and function the same way as physical ECUs. The use of virtual ECUs (vECUs) helps automakers to achieve faster product development while reducing the need to procure physical components, including equivalent physical ECUs and wiring looms and related components. Moreover, vECUs enable testing and development teams to more easily conduct comprehensive testing across varied environments.
vECUs accelerate development and early-stage validation by enabling the simulation of diverse driving conditions, system failures, and edge cases that are difficult to replicate in physical testing. They also support collaborative development across geographically distributed teams, facilitate the creation of unified software platforms across different vehicle models, and improve safety and cybersecurity by allowing vulnerabilities and attack scenarios to be identified, tested and addressed before deployment.
Some of the leading companies that are offering and deploying vECUs are dSPACE VEOS, Continental and BMW.
dSPACE VEOS enables ECU software to be simulated and tested on standard PCs in a Software-in-the-Loop environment, allowing engineers to run virtual ECUs and full vehicle models early in development without relying on physical hardware.
Continental is already deploying virtual ECUs through its vECU Creator platform, developed in collaboration with AWS. With vECU Creator, OEMs, suppliers, and third-party developers can configure and run cloud-based vECUs. This allows software development, testing, and validation to begin before the underlying microcontroller, processor, or high-performance computing hardware is physically available.
BMW also uses vECUs to decouple software development from hardware, enabling early simulation, testing, and validation of vehicle software. This reduces development time while maintaining safety and quality and accelerates software deployment in modern centralised and zonal vehicle architectures.
As vehicles evolve into software-defined systems, their computing demands increase dramatically. Users now have more expectations from connected vehicles, including supporting advanced driver assistance systems, autonomous driving capabilities, and infotainment platforms. Instead of relying on single, monolithic chips to handle all processing needs, the industry is moving toward modular chiplet-based architectures. With this approach, different specialised semiconductor chips are combined into a unified package, allowing manufacturers to mix and match components based on specific functional requirements.
Using chiplets improves flexibility, reduces redundancy, and helps avoid vendor lock-in through the use of standardised, interchangeable components. As automakers consolidate workloads onto high-performance central platforms that manage multiple vehicle domains, these platforms must integrate GPUs, NPUs, and AI accelerators to process sensor data, enable real-time decision-making, and support increasingly complex AI workloads. The use of chiplets helps OEMs to achieve this flexibly and efficiently.
Some of the leading companies that are working on chiplets are discussed below.
Cadence and Arm partnered to design a standard chiplet reference blueprint, developed for advanced driver assistance systems (ADAS). This was an initiative intended to eliminate interoperability issues and to foster an ecosystem for collaboration between multiple chiplet vendors.
IMEC established the Automotive Chiplet Programme (ACP) by partnering with industry players from across the automotive ecosystem, including Arm, BMW Group, Bosch, Cadence, Siemens, SiliconAuto, Synopsys, and Tenstorrent. An objective of the program is to stimulate the development of an open chiplet ecosystem to deliver automotive compute and safety requirements.
BOS and Tenstorrent introduced Eagle-N, an automotive AI accelerator chiplet system-on-chip. The chiplet SoC is integrated with Tenstorrent’s NPU core, supporting various automotive applications such as ADAS and in-vehicle infotainment.
Intel announced the Intel SDV SoC, which uses a multi-node chiplet architecture to deliver advanced AI-driven performance in vehicles.
The shift toward SDVs represents a fundamental rethinking of automotive engineering. Vehicles are no longer limited solely by hardware innovations. Instead, they are becoming dynamic computing platforms that evolve continuously through software. The three building blocks discussed in this blog are not independent developments. Rather, they are closely interconnected elements of a broader industry transformation. Together, these innovations are not just improving vehicles; they are redefining what a vehicle is.
