LITEON and NVIDIA Partner on AI-RAN for 5G/6G
At MWC Barcelona 2026, LITEON Technology announced it is accelerating AI-RAN commercialization by integrating its radio units with NVIDIA's AI Aerial platform. The partnership focuses on using GPU-accelerated architectures to create scalable, AI-native Radio Access Networks. This collaboration aims to improve the performance and efficiency of next-generation wireless infrastructure.
The partnership between LITEON and NVIDIA centers on a technology called AI-RAN, which integrates AI capabilities directly into the radio access network. This approach moves beyond traditional network architectures by using general-purpose GPUs to handle both cellular and AI workloads on the same infrastructure. The goal is to improve spectral efficiency, dynamically manage traffic, and enable new AI-based services at the network edge. This collaboration leverages NVIDIA's AI Aerial platform, a suite of software and tools for building AI-native wireless networks. LITEON is integrating its Open RAN-compliant radio units (RUs) with this platform. A key aspect of LITEON's RU architecture is its support for the O-RAN 7.2x fronthaul interface, which standardizes communication and reduces integration complexity for operators. The move towards AI-RAN is a strategic shift from Virtualized RAN (vRAN) and Open RAN. While vRAN decouples network software from proprietary hardware and Open RAN promotes interoperability between vendors, AI-RAN embeds AI and machine learning directly into the radio signal processing layer. This allows for real-time network optimization and the deployment of demanding AI applications directly at the network edge. NVIDIA's broader strategy involves creating a full stack for AI in telecommunications, working with partners like Nokia, Cisco, and T-Mobile. This includes the development of platforms like the Aerial RAN Computer Pro (ARC-Pro) and leveraging their Grace Hopper Superchips, which combine CPU and GPU capabilities to efficiently process large datasets for both 5G/6G and AI applications. At MWC Barcelona 2026, LITEON is demonstrating this integration alongside partners SynaXG and Supermicro. One demonstration by SynaXG showcased a fully software-defined AI-RAN on a single NVIDIA GH200 server, achieving a throughput of 36 Gbps with latency under 10 milliseconds, a first for millimeter-wave (FR2) bands. This technology is seen as a foundational step toward 6G, which is expected to be AI-native from the ground up. Future 6G networks will need to support a massive number of AI-driven devices like robots, autonomous vehicles, and sensors. The AI-RAN architecture allows operators to upgrade from 5G to 6G primarily through software, running both generations simultaneously on the same hardware. LITEON has been an active participant in the O-RAN Alliance, focusing on developing open and interoperable 5G small cell solutions for private networks in environments like smart factories. Their work with Anritsu on performance testing and previous showcases at MWC have highlighted their focus on energy-efficient and O-RAN compliant hardware. The industry-wide push for AI-RAN is formalized by the AI-RAN Alliance, of which both LITEON and NVIDIA are a part. This group aims to shape the future of AI-native networks by creating a functional framework for integrating AI directly into the RAN infrastructure. The alliance's work is expected to accelerate the global commercialization of these technologies ahead of the traditional 6G deployment cycle.
