The most dangerous misconception in this automotive software race is thinking that adding more code simply makes a car “software-defined.” Sonatus’ leadership insists that the real transformation hinges on a tightly integrated loop of perception, decision-making, and execution – a now-evolving architecture into the AI-Defined Vehicle, AIDV, era.
To Yu Fang, Co-Founder and CTO of Sonatus, perception means equipping OEMs with accurate, real-world operational data on both vehicle systems and customer usage patterns. Then, decision-making is the analytics layer that transforms that data into actionable insights, while execution is the rapid deployment of those insights into live vehicle functions. This closed-loop approach redefines value creation: in the SDV era, the moment of delivery marks the start-not the end-of a vehicle’s evolution.
To that end, Sonatus has developed a suite of foundational products focused on Software-Defined Networking (SDN), Software-Defined Storage (SDS), and Software-Defined Compute (SDC). SDN enables the dynamic reconfiguration of in-vehicle network paths independent of firmware overhauls, leveraging automotive Ethernet to reduce wiring weight and simplify diagnostics. SDS enables cross-domain sharing of data, tearing down ECU silos by allowing applications to acquire stored information without complex real-time communication. SDC, essential for centralized computing platforms, leverages containerized application management to isolate workloads, allocate compute resources with precision, and maintain performance integrity-vital as vehicles consolidate dozens of ECUs into high-performance domain controllers.
These functionalities are also in line with industrial movements toward centralized computing platforms and the deployment of containerized applications, which require cross-domain functionalities and OTA updates. Supported by firmware, container, configuration, and hybrid updates, the multi-modal OTA system from Sonatus allows for true iterative development at the vehicle level.
The leap from SDV to AIDV is driven by integrating AI into that infrastructure. “SDV is the infrastructure; AI is the brain,” Fang says. Without SDV’s robust data acquisition, governance, and execution layers, deploying AI at the vehicle edge is impossible. Wallie Leung, SVP of Sales & Business Development, says AI’s role goes beyond in-vehicle personalization or predictive maintenance to accelerating design, simulation, and testing cycles. Collector AI and Automator AI illustrate this synergy: Collector AI utilizes trigger-based strategies to enhance data quality, while Automator AI enables no-code function deployment, reducing time-to-market. AI Director addresses one of the most challenging problems: model deployment on heterogeneous vehicle platforms, while AI Technician applies predictive diagnostics to reduce warranty and recall costs.
This approach solves for key technical pain points associated with the deployment of vehicle-edge AI: latency, model size constraints, and OTA update complexity. In treating AI as a shared service across the vehicle rather than embedding it in individual applications, Sonatus reduces software bloat and improves efficiency-a philosophy in sync with the emerging full-domain operating systems in China’s automotive sector.
Ecosystem collaboration is core to scaling both SDV and AIDV. Sonatus teams up with hardware and chipmakers, such as suppliers of high-performance automotive SoCs and NPUs, to ensure compatibility and best performance. To date, integration with hyperscale cloud providers such as Google, AWS, and Microsoft features seamless vehicle-cloud data exchange for hybrid AI architectures that balance edge and cloud processing. Co-creation with AI model companies enables Sonatus to focus on deployment and optimization rather than building large models from scratch, thus avoiding redundancy in the ecosystem.
That openness is underpinned by standards compliance. Sonatus supports AUTOSAR and VSS, with software running on POSIX-compliant OSs to maximize portability. Flexible SDKs and event APIs simplify integration, enabling OEMs to reuse software across platforms and reduce development costs key in a market where cross-domain data sharing and abstraction layers are turning out to be strategic differentiators. China’s rapid cycle of adoption provides a proving ground that showcases the technologies being taken up by Sonatus. Solutions are iterated by local OEMs in six to twelve months, showcasing Silicon Valley-like agility.
The China team at Sonatus participates in both POC and SOP projects. Early signals for the arising need enter into the global roadmap. Chinese automakers are expanding overseas, making Sonatus’ experience in data compliance, localization, and adaptation across regions a crucial asset. It is a self-sustaining feedback loop where local innovation informs global deployment and vice versa.
By integrating the power of SDV’s infrastructure with adaptive intelligence from AI, Sonatus positions vehicles for autonomous, self-optimizing systems. This vision demands mastery of both centralized compute architectures and edge AI execution, supported by a strong ecosystem and adherence to global standards. To OEMs and industry innovators, the message is clear: competitive advantage in next-generation mobility will come from integrating the software foundation with the AI brain-seamlessly, securely, and at scale.
