Nokia has announced new AI-RAN capabilities with NVIDIA and launched a sovereign agentic AI platform for telco autonomy across network operations. The capabilities apply AI within radio processing, support intelligence close to where data is generated and extend governed AI agents across network operations.
Building on the industry’s first commercial AI-RAN platform, announced in July, the companies say AI-RAN is being integrated into live networks and moving closer to deployment. Nokia and NVIDIA are developing a real-time network foundation for physical-AI applications, while providing operators a path to improve 5G and 5G Advanced networks and evolve to AI-native 6G.
Nokia AI-native anyRAN software, running on NVIDIA Aerial RAN Computer, provides a software-defined foundation for AI-powered radio functions. The companies are applying AI within the 5G signal-processing chain, including channel estimation, beamforming, multi-user MIMO pairing and scheduling, to improve network capacity and performance.
Customer programs include Elisa’s demonstration in Helsinki, where voice, mobile data and video analytics ran simultaneously on the same AI-RAN platform in its 5G network. Distributed deployment and evaluation of customer use cases and commercial AI-RAN services are expected in 2027. BT is preparing a live-network trial in Birmingham in 2027, while Orange plans testing in lab and real network conditions. Nokia and Deutsche Telekom are developing AI-based channel estimation; initial laboratory measurements showed typical uplink throughput improvements of 10–15%, with greater gains under more complex radio conditions.
Nokia and NVIDIA are integrating NVIDIA AI Enterprise software and NVIDIA Nemotron models and libraries into Nokia’s agentic platform. Operators will be able to build, evaluate, deploy and operate telco-specific agents across radio, core, transport and cloud networks. Nokia provides network intelligence, operational context and governance; glass-box autonomy maintains operator oversight, while agentic guardrails govern how actions are tested and executed. The architecture supports on-premises, cloud and hybrid deployments, giving operators control over where operational data, models and policies reside.
Nokia is also extending the AI-RAN ecosystem through distributed applications (dApps) that run at the network edge. Nokia’s E3 interface gives the applications real-time access to RAN data. DeepSig is developing spectrum-sensing software to identify RF threats, interference and rogue transmitters, while ZaiNar is integrating positioning software with Nokia Cloud AI-RAN for network-based location capabilities using existing wireless infrastructure. At NVIDIA GTC Berlin, Nokia is demonstrating integrated sensing and communications and showcasing IP and optical solutions for distributed AI infrastructure.
