LITEON Taps NVIDIA for AI-RAN
LITEON is accelerating its push into AI-native radio access networks (AI-RAN) by integrating its hardware with NVIDIA's AI Aerial platform. The company is showcasing the GPU-accelerated architecture at MWC Barcelona, aiming to speed up the commercialization of AI-powered 5G infrastructure.
This partnership places LITEON's hardware within the NVIDIA AI Aerial ecosystem, a platform designed to merge AI and cellular network operations onto a single, GPU-powered infrastructure. The goal is to replace specialized, single-purpose telecom hardware with flexible, software-defined systems that can handle both 5G traffic and demanding AI applications. NVIDIA's Aerial platform provides the software framework for this shift, including CUDA-accelerated libraries and AI models that can improve network performance. It also offers a "digital twin" platform using NVIDIA Omniverse, allowing for the simulation and testing of entire city-scale 5G and 6G networks before deployment. LITEON's contribution is the critical hardware: O-RAN compliant 5G Open Radio Units (O-RUs). As a member of the O-RAN Alliance, LITEON develops hardware based on open and interoperable standards, which helps telecom operators avoid vendor lock-in and allows for more flexible and cost-effective network rollouts. The integration of AI into the radio access network (AI-RAN) can significantly boost network performance. Tests have shown that leveraging AI can improve spectral efficiency by up to 20%, allowing more data to be transmitted over the same bandwidth. This unified infrastructure also enables dynamic allocation of resources, potentially increasing capacity utilization by two to three times. This technology unlocks new revenue opportunities for network operators by turning cell sites into distributed AI computing engines. This "AI on RAN" model can power low-latency edge services like autonomous robots, smart factory applications, real-time video analytics, and AR/VR experiences. The collaboration is a key step towards future 6G networks, which are expected to be AI-native from the ground up. By creating a software-upgradable path, operators can prepare for 6G without costly and disruptive hardware replacements, positioning the network to handle future AI-driven demands.
