The semiconductor paradox in today’s cars

Today’s vehicles require semiconductors that can be easily assembled like a universal toolkit and deployed across the entire E/E architecture, yet still be configured to such an extent that they meet an OEM’s specific software, safety and efficiency requirement. This apparent paradox is resolved when semiconductors are no longer viewed as isolated single purpose devices but as modular platforms that combine different functional and performance profiles, along with matching safety concepts, on a common technology foundation.

New vehicle E/E architectures

Vehicle E/E architecture is evolving from various narrowly focused applications towards zonal and domain-based concepts with central high-performance computers and powerful zonal nodes featuring edge AI capabilities. Instead of equipping every function, such as window lifters, seat adjustment, lighting, radar or steering actuators, with a dedicated custom chip, OEMs increasingly rely on broadly applicable MCUs and SoCs that support multiple tasks and are configured via software (Figure 1).

This architecture follows the principle of “central intelligence, distributed compute”: central computers handle complex ADAS, infotainment and vehicle functions, while zonal controllers act as gateways, sensor hubs and actuator controllers in each vehicle while also performing initial local processing and evaluation steps. As a result, the number of ECUs and wiring harnesses decreases, while data rates, safety requirements and the number of software functions rise significantly – a combination that is impossible to manage without semiconductors that are both widely deployable and highly adaptable.

For semiconductor strategies, this means OEMs and Tier 1 suppliers are moving to chip families that can play different roles – from simple body controllers to safety critical ECUs. Tailored solutions arise less from a multitude of completely different ASICs and more from scalable platforms offering variants in memory size, compute performance, interfaces, safety level and packaging. Such platforms form a continuous “silicon backbone” that spans multiple product generations and via which new vehicle lines, derivatives and special editions can be launched more quickly.

Subscription models, performance headroom and energy demand

With software first and subscription-based business models, vehicles are turning into a versatile, software-defined platform where hardware-ready features can be unlocked, expanded or adapted via licenses over the entire vehicle lifetime. Even in entry-level segments, this requires semiconductors that offer significantly more performance than initially used. It includes, for example, supporting additional comfort features, higher levels of automation or new diagnostic services that can be rolled out even years after the vehicle has already been shipped. This places considerable demands on the performance and, above all, the reliability of the semiconductors (Figure 2).

At the same time, this performance headroom must not become a permanent load on the onboard power supply, especially in electric vehicles with long parking and monitoring phases. Always on functions such as environmental sensing, telematics, OTA connectivity and safety monitoring increase the energy consumption of the electronics, while range and overall efficiency remain key metrics. Semiconductor platforms therefore need to support advanced energy management concepts – from deep-sleep modes and dynamic voltage and frequency scaling to intelligent workload distribution within the devices, for example between cores and peripheral blocks.

On the system level, intelligent strategies are also required: for instance, shifting computationally demanding tasks into short, clearly defined activity windows while allowing large parts of the hardware to remain in highly efficient low power states most of the time. In addition, mechanisms such as graceful degradation, adaptive quality levels for sensors, and data driven optimization of energy consumption are gaining importance to continuously improve the interaction between software functions and semiconductor platforms in the field.

What challenges will the wiring harness industry face over the next five years?

Which challenges will shape the wiring harness and EDS industry most strongly over the next five years? Four experts – all speakers at the 2026 Bordnetzkongress in Germany – provide surprisingly different answers.

Benjamin Müller

At the same time, sustainability pressure is increasing. OEMs expect semiconductor solutions that are not only efficient in operation but also offer high yield, long-term availability and robust reliability to minimize retrofits, spare part diversity and premature obsolescence. Universal platform chips can contribute to this by driving higher manufacturing volumes and longer lifecycles, which in turn improve fab efficiency and simplify variant management in the field. From the semiconductor manufacturer’s perspective, this approach also offers advantages: It reduces product variants while increasing the production volume of platform components which unlocks significant scale effects. Process control in manufacturing relies on statistical evaluation of production data, which becomes more robust at higher volumes, thereby simplifying process control. In practice, this typically translates into higher yields and better quality that can be achieved more cost effectively and efficiently.

Technology levers for universal chip platforms

To partition software from hardware, balance performance, energy efficiency and scalability, modern automotive semiconductors increasingly rely on process technologies that support digital high performance logic, mixed signal, RF, mmWave, NVM and power functionality within a single node. FD SOI based processes are particularly relevant here, as the isolating substrate layer and body bias techniques enable operation at very low voltages and deliver significant advantages in leakage and dynamic power compared to conventional bulk CMOS (Figure 3).

This allows for the development of MCUs that provide high performance in active mode while remaining available for long periods in standby with minimal energy consumption – a crucial capability for zonal controllers or continuous vehicle monitoring. At the same time, these technologies enable integration of embedded non-volatile memories (e.g., eFlash, MRAM, RRAM), robust high voltage drivers and RF/mmWave blocks for radar or connectivity on a single chip, reducing system complexity, component count and overall cost.

Another lever is the systematic use of design platforms: reusable IP libraries for CPU cores, peripheral blocks, security modules, network interfaces and safety mechanisms make it possible to derive multiple device families for different in vehicle tasks from the same technology. This keeps mask costs and development times under control, while OEM specific differentiation continues to be achieved via configuration, IP selection and software implementation.

In addition, clearly defined automotive process technologies with guaranteed quality levels, mission profile qualified temperature ranges and adapted supply models such as turnkey (wafer manufacturing plus packaging and test) can improve quality, reliability and long term availability. This significantly reduces product risk already early in the project phase and accelerates industrialization of new derivatives, as qualification, reliability data and supply commitments are aligned with automotive lifecycles and operating conditions.

Sponsored

Europe Automotive Computing Conference

Munich on November 18-19, 2026

How “custom” is being redefined

Historically, “custom silicon” meant developing a dedicated, highly specialized chip for a specific function with precisely tailored peripherals, memory size and interfaces. Today, that siloed development model has hit a breaking point; it simply cannot keep pace with the hundreds of concurrent functions and rapid update cycles required by modern vehicle platforms

Today, the notion of “custom” is shifting from fixed hardware toward programable flexibility: a semiconductor platform is considered custom tailored when it offers a modular base architecture that can be precisely adapted to different OEM E/E architectures, vehicle segments and functional scopes via variants and configuration. This includes options for scaling CPU performance, different memory and I/O configurations, safety concepts up to ASIL D, differentiated security features and dedicated IP blocks for radar, motor control or power distribution.

Under this new definition, 'custom' means the hardware remains constant while the vehicle’s identity evolves through software. By moving away from bespoke, one-off chips in favor of a standardized technology platform, semiconductor suppliers offer a versatile 'silicon backbone' for the E/E architecture. This approach eliminates supply chain headaches, as fewer distinct part numbers need to be managed and qualified over the long term (Figure 4).

For Tier 1 suppliers, this new form of customization opens additional degrees of flexibility: they can build diverse ECU portfolios on top of an established platform, with differentiation increasingly driven by software, system integration and application know how rather than by fundamentally different silicon designs. As a result, competition shifts towards system architecture, software quality and end user experience, while the semiconductor platform retains a stable, industrially optimized foundation.

Enabler for software first strategies

Today’s software defined vehicles require hardware to be treated as a scalable, homogeneous infrastructure on which functions can be provided, orchestrated and updated as services. Versatile and flexible chip platforms create the conditions for this by offering standardized compute and communication resources in zonal controllers and central high performance computers that operate independently of specific vehicle variants.

Abstraction of physical inputs and outputs into software-defined services, such as “front left window lifter”, “front camera” or “front center radar”, only becomes efficient when the underlying MCUs and SoCs provide largely identical software interfaces, security mechanisms and diagnostic capabilities. Under these conditions, feature updates, function bundles or new subscription offerings can be rolled out via OTA across large fleets without having to maintain separate software branches for every vehicle line or ECU type.

For OEMs, this means that innovation and go-to-market speed is increasingly determined by software teams, while the semiconductor platform quietly ensures continuity, scalability and reliability in the background. The better safety, security, real time capability, energy efficiency and cost targets are already embodied in this platform, the more smoothly software first strategies can be implemented and the easier it becomes to resolve the semiconductor paradox in the modern car.

WeRide and Uber bring Robotaxis to Madrid

WeRide and Uber are taking their Robotaxi partnership to Europe. Madrid will host the first commercial pilot, offering a glimpse into how autonomous ride-hailing could scale across regulated urban markets.

Benjamin Müller