LITEON Technology announced a major milestone in AI-RAN innovation at COMPUTEX 2026 through its collaboration with Singapore University of Technology (SUTD) and SUTD spin-off company NeuroRAN.
This joint demonstration will show how AI can be natively integrated into 5G radio access networks (RANs) to enable real-time, privacy-preserving, and energy-efficient edge AI applications.
At the core of the solution is dynamic AI model segmentation. LITEON will provide high-performance O-RUs (radio units) integrated with AI-RAN infrastructure, while SUTD will develop AI applications leveraging advanced deep learning models such as Swin Transformer for real-time video object detection. The solution runs on NVIDIA AI Infrastructure and integrates NVIDIA AI Aerial software with high-performance computing to tightly integrate AI with 5G networks.
A key innovation is the use of AI-driven spectrum sensing to inform decisions to dynamically distribute AI workloads between user equipment (UE) and edge/cloud environments based on real-time network conditions. Compared to traditional fixed partitioning, this approach significantly reduces latency, optimizes energy consumption, and ensures that sensitive data remains on-device, enhancing both privacy and performance.
The demonstration also integrates distributed UPF (dUPF) and 5G core networks to further reduce end-to-end latency and enhance edge AI deployment capabilities for real-world applications such as smart cities, intelligent surveillance, and industrial automation.
LITEON continues to expand its AI-RAN ecosystem partnerships to accelerate the deployment of real-time AI applications. In addition to commercializing AI-RAN, LITEON plans to provide carriers with a flexible AI development environment powered by the NVIDIA DGX Spark platform, enabling carriers to accelerate research, validation, and deployment of next-generation edge AI applications and services. By leveraging DGX Spark for AI model development and small-scale testing, carriers can quickly prototype innovative AI-powered use cases before scaling across commercial 5G and future network infrastructure.
Richard Chiang, General Manager, LITEON Smart Life Application Strategy Business Unit
Through our collaboration with SUTD and NeuroRAN, we are accelerating AI-RAN from vision to commercialization. By combining LITEON’s expertise in O-RAN and wireless infrastructure with advanced NVIDIA AI platforms and applications, we aim to provide carriers and enterprises with more flexible and efficient network architectures.
Chen Bin Bin, SUTD Associate Professor, Future Communications Research and Development Program (FCP) Deputy Director
SUTD and our spin-off company NeuroRAN are excited to partner with LITEON in the development of this cutting-edge AI-RAN demo. This shows how intelligent 5G networks can play a key role in enabling new physical AI use cases.
