LITEON and NVIDIA Accelerate AI-RAN at MWC
At MWC 2026, LITEON is showcasing how it's accelerating AI-RAN commercialization by integrating its Open Radio Units with NVIDIA's AI Aerial platform. The collaboration highlights how GPU-accelerated architectures are enabling the telecom industry's shift to AI-native networks.
The underlying technology, NVIDIA's Aerial platform, is a suite of software and hardware designed to unify Radio Access Network (RAN) and AI workloads on a single GPU-powered infrastructure. This allows for dynamic allocation of resources based on real-time demand, which can increase capacity utilization by two to three times compared to siloed systems. The platform is designed to be a future-proof pathway to 6G, enabling network transitions through software upgrades rather than costly hardware replacements. This collaboration is a key part of the AI-RAN Alliance, a group formed in February 2024 with founding members like NVIDIA, Samsung, Arm, Microsoft, and T-Mobile. The alliance aims to integrate AI into cellular technology to boost network efficiency and enable new AI-driven revenue opportunities by deploying AI services at the network edge. At MWC 2026, the now 132-member alliance is showcasing 33 AI-driven demonstrations and four new industry blueprints. LITEON's contribution centers on its Open Radio Unit (O-RU) architecture, which provides the carrier-grade stability and consistent performance necessary for AI-RAN systems. These O-RUs support the O-RAN 7.2x fronthaul interface, which helps reduce integration complexity for operators. LITEON has already completed integration of its sub-6 GHz and millimeter wave radio units with the NVIDIA Aerial platform. For ML engineers, this trend signifies a major shift towards deploying AI at the network edge. AI-native networks are designed to be autonomous, using continuous feedback loops and predictive optimization to manage resources, detect threats, and reduce latency for AI workloads. This creates a need for engineers skilled in MLOps, production deployment, data engineering for real-time pipelines, and building AI models for functions like traffic prediction and anomaly detection. The move to AI-RAN is a response to the steep decline in the traditional RAN market, which fell by nearly $9 billion in 2024 from its 2021 peak. Analysts project the AI-RAN market will grow significantly, with some estimates predicting a rise from $1.7 billion in 2025 to $10.4 billion by 2030. While early deployments have already shown up to a 20% increase in downlink throughput and a 15-20% reduction in energy use, large-scale commercial rollouts are not expected before 2026, with broader adoption likely by 2027. Beyond network optimization, running AI and RAN workloads on the same infrastructure opens new revenue streams for telecom operators. This shared infrastructure can host AI applications like intelligent video analytics, industrial automation, and generative AI services directly at the edge. Field trials are already underway, with T-Mobile demonstrating concurrent AI and RAN processing and SoftBank achieving a 16-layer massive MIMO using a fully software-defined 5G on NVIDIA's platform.
