LITEON and NVIDIA Push AI-Native 5G
At MWC Barcelona, LITEON Technology showcased its new AI-native Radio Access Network (RAN) architectures. By integrating open radio units with NVIDIA's AI Aerial platform, the company is accelerating the shift from hardware-centric telecom networks to scalable, software-defined, GPU-driven systems.
The move to AI-native RAN is a strategic shift to handle the exponential growth in network traffic driven by AI applications. This new architecture allows for the dynamic allocation of 5G and AI workloads on the same GPU, which can increase capacity utilization by two to three times. For enterprises, this means the network itself becomes a platform for low-latency AI inference at the edge, opening doors for new intelligent applications in logistics, manufacturing, and retail. LITEON's O-RAN compliant 7.2 split 5G open-radio units (O-RU) are a key component, designed for high performance and energy efficiency. This open architecture is crucial as it prevents vendor lock-in and allows operators to mix and match hardware and software from different providers, fostering innovation and reducing costs. The collaboration aims to ensure LITEON's radio units are fully compatible with NVIDIA's platform, reducing integration complexity for faster deployment. The NVIDIA Aerial platform provides a software-defined, accelerated computing suite for developing and deploying these AI-native wireless networks. It includes tools like the Aerial Omniverse Digital Twin for physically accurate simulations of wireless systems, from a single cell to city-scale environments. This allows for extensive testing and optimization before commercial rollout. This collaboration is part of a larger industry trend, with an AI-RAN Alliance forming to guide the development of these next-generation networks. Major telecom operators like T-Mobile, SoftBank, and Nokia are already conducting live field trials of AI-RAN systems powered by NVIDIA's technology. These trials are demonstrating the real-world viability of running concurrent AI and RAN processing on the same infrastructure.
