The transition to Software-Defined Vehicles (SDVs) goes beyond ECU consolidation. It represents a shift toward redesigning the entire vehicle architecture around software. As software-defined capabilities expand from the cockpit and ADAS to body, chassis, and safety control, a foundation that enables different operating systems and requirements to coexist safely on shared hardware becomes increasingly important.

This session explores SDV architecture spanning AI computing and safety MCUs, and explains the significance of hardware–software decoupling. It examines the role of the hypervisor in supporting functional safety, cybersecurity, and real-time performance in environments where Android, Linux, AUTOSAR Classic, RTOS, and AI workloads coexist.

The session reviews evolving industry architectures through multidomain integration trends in high-performance SoCs such as NVIDIA Thor and Renesas R-Car X5H, alongside virtualization approaches using Android Automotive and VirtIO. It then introduces the CPU and MCU architecture of PEGASUS, an ISO 26262 ASIL-D certified hypervisor, including hardware resource partitioning and isolation techniques that help contain faults and cyber threats across operating systems. It also presents the achievement of Automotive SPICE® v4.0 Capability Level 2 for the development processes of the GAIA secure bootloader.

A NVIDIA Thor-based multi-OS demonstration illustrates the virtualization configuration on real hardware. The session also explains how supported platforms and evaluation environments can be used to begin validating an SDV architecture.

Designed for OEM, Tier 1, and Tier 2 engineers and technology planning professionals working on automotive software and E/E architectures, this session provides design perspectives for safely integrating diverse software and reducing hardware dependencies.