LITEON and NVIDIA Partner on AI-Native RAN
At MWC Barcelona, LITEON Technology announced it is integrating its Open Radio Units with NVIDIA's AI Aerial platform. The partnership aims to accelerate the commercialization of AI-RAN (Radio Access Network). This move highlights the industry's shift toward GPU-accelerated, AI-native architectures for next-generation telecommunications.
The NVIDIA Aerial platform is a comprehensive suite for building AI-native wireless networks, providing the necessary hardware and software to move from prototyping to commercial deployment of AI-RAN solutions for 5G and 6G. This shift to a software-defined, GPU-accelerated architecture allows for the dynamic allocation of both 5G and AI workloads on the same hardware, which can increase capacity utilization by 2-3 times. LITEON has integrated its sub-6 GHz and millimeter wave radio units with this platform, aiming to reduce integration complexity and speed up deployment. This collaboration is part of a larger industry trend preparing for 6G, which is expected to be an AI-native platform from its inception. The AI-RAN Alliance, an industry group, is working to advance these AI-driven infrastructures. By embedding AI into every layer of the network, future RANs are expected to be self-evolving and capable of autonomous learning and optimization. A recent industry survey indicated that 77% of respondents foresee a much faster deployment timeline for this new AI-native architecture ahead of the typical 6G cycle. The core of NVIDIA's platform includes the Aerial CUDA-Accelerated RAN, a software development kit for building 5G and 6G base station software on NVIDIA's accelerated computing platforms. This leverages massive GPU parallelism to accelerate critical tasks like signal processing, scheduling, and beamforming. Hardware solutions like the NVIDIA GH200 Grace Hopper Superchip combine CPU and GPU capabilities to handle the demanding, co-located processing of large datasets and AI workloads. For operators, the primary benefits of applying AI to the RAN are improved performance and operational efficiency. AI can enhance spectral efficiency, optimize traffic flows, and automate network management. This allows for the support of more connections with better performance. LITEON is also partnering with companies like SynaXG and Supermicro to create a unified architecture for running RAN and AI workloads, enabling low-latency AI inference at the network edge.
