Visteon has developed an edge-to-cloud AI arbitration architecture that dynamically allocates workloads between on-device hardware and cloud infrastructure. Runtime intelligence routes latency-sensitive functions to the vehicle edge while offloading complex reasoning tasks like fleet learning and predictive maintenance to the cloud.

Building on the AI-ADAS Compute Module showcased at CES, the platform leverages NVIDIA DRIVE AGX Orin, DriveOS software and NVIDIA open-source TensorRT-Edge-LLM SDK. Generative AI capabilities are enhanced through NVIDIA Nemotron open models and NIM microservices.

The production-ready solution bundles AI model training, optimization and Visteon's automotive-grade AI operating system (AIOS) stack. It supports data residency controls, regulatory compliance and global scale, enabling automakers to deploy agent-centric in-cabin assistants and other AI-driven features across vehicle platforms.