The solution addresses a critical challenge for operators: traditional approaches require separate systems for the radio access network and AI applications, which can double capital costs and create significant integration complexity.
By combining AMD EPYC™ processors with the Wind River Cloud Platform, the company offers a production-ready solution capable of hosting virtualized RAN functions and AI inference workloads in parallel on the same distributed platform. This architectural approach allows operators to introduce real-time AI capabilities—such as traffic prediction, anomaly detection, energy optimization, and network intelligence—without the infrastructure duplication that has previously limited AI-RAN adoption.
The jointly developed platform enables operators to:
Reduce infrastructure costs and operational complexity by unifying Open RAN and AI-RAN workloads on shared hardware.
Deploy AI-based capabilities directly at the network edge, including real-time traffic prediction, anomaly detection, and power optimization alongside virtualized RAN functions.
Scale efficiently across thousands of distributed sites with automated lifecycle management, resilience, and fault tolerance.
Evolve without hardware replacement, adding advanced AI capabilities as needs grow, while preserving architectural flexibility.
Javed Khan, executive vice president and president of Intelligent Systems at Aptiv, noted that as operators move from Open RAN trials to commercial deployments, AI becomes a critical capability rather than a future add-on, enabling the incorporation of intelligence without duplicating infrastructure.
Philip Guido, AMD's Chief Commercial Officer, emphasized that AI solutions for telecommunications require a flexible, high-performance computing foundation capable of supporting real-time RAN workloads and AI inference. He added that AMD EPYC processors offer the performance and scalability needed for AI-driven RAN architectures.
The alliance includes joint engineering focused on optimizing the AI-RAN software stack and hardware-software co-optimization, a development roadmap, and proof-of-concept deployments.
