Last month, Yole Group forecast a $160 billion automotive semiconductor market by 2031 with a 31% CAGR between 2025 and 2031, in its Automotive Semiconductor Trends 2026 report. That growth is increasingly driven by the number of chips per car rather than higher value content and rebuilding the car’s electrical and electronic (E/E) architecture is one of the main forces driving demand for that new silicon.
Yole Group now expands on this shift with a dedicated new report, Automotive E/E Architectures 2026. The analysis zooms in on the transition from over 100 scattered control units to a handful of domain, zonal, and central computers, the rise of the software-defined vehicle (SDV), and the specific processors, network chips, and power devices powering this evolution.
Decades of complexity being dismantled
The network of controllers, processors, communication chips, and power devices that make up a car’s E/E architecture is undergoing its biggest change in decades. For years, the wiring harness was one of the heaviest and most expensive subsystems on board, as carmakers stacked function-specific ECUs until a single vehicle could carry over 100 units and kilometres of cable.
That model is being dismantled. Driven by the SDV, the rise of ADAS and AI, and the pressure to cut cost and weight, carmakers are consolidating scattered ECUs into a few high-performance domain, central, and zonal controllers. The benefits are already measurable: Tesla cut the wiring for the Model 3 to half that of the Model S, Rivian reduced its ECU count from 17 to seven, and BMW’s Neue Klasse saves around 600 meters of cable while replacing up to 150 fuses with smart e-fuses.
The Automotive E/E Architectures 2026 report quantifies who wins, when, and by how much, mapping the transformation end-to-end, from architectural strategy down to the individual chip.
“This E/EA transition is creating new demand across the automotive semiconductor stack: advanced processors for central AI-based computation ($5.6 to $15.3 billion, 16% CAGR), high-performance MCUs for domain and zonal controllers (6.6% CAGR), in-vehicle network ICs as CAN/LIN give way to CAN FD and Ethernet (doubling from $3.9 to $8.1 billion), and a nearly new category of PMICs for 48V powernets (173% CAGR),” said Pierrick Boulay, Principal Analyst, Automotive Semiconductors at Yole Group.
Focusing on light vehicles with forecasts spanning 2021-2031, Yole Group’s report tracks the architectural shift from distributed to domain, zonal, and central designs, including one-box, one-board, and one-chip integration. It shows how the pace of change diverges sharply by region: Chinese OEMs are set to reach 56% zonal production by 2031, compared with just 18% outside China.
“In the long-term trend of centralised E/E architectures, we are in the phase of upgrading backbone networks from domain-based to cross-domain zonal architectures plus a central computer for ADAS, infotainment, and vehicle control. Many new technologies are on the horizon, creating new opportunities for semiconductor devices: e-fuses, 48V powernet, optical communication, etc.,” said Yu Yang, Principal Lead Analyst, Automotive & Robotics at Yole Group.
The report also maps the semiconductor landscape enabling this transition, including automotive processors and the RISC-V ecosystem, microcontrollers, network chips, and power management ICs. It further examines how OEMs and Tier-1s are sourcing this silicon, weighing the make, buy, or partner decision that keeps full in-house design within reach of only high-volume players like Tesla and a handful of Chinese OEMs.
