AI for RAN

AI for RAN enables leveraging AI to enhance RAN performance and efficiency. This is possible whether the related AI workloads run within the base station or in data centers. 

Mobile network operators see huge potential in achieving further efficiency and performance gains in 5G networks with AI for RAN. AI is vital for achieving autonomous RAN operation levels 4 and 5 according to TM Forum’s autonomous network framework. 

6G will be AI-native. Nokia demonstrated an AI-native air interface in 2023, proving its feasibility and capabilities. 

With a comprehensive Service Management and Orchestration (SMO) solution, operators can benefit from open innovation facilitated by AI-powered radio applications (rApps).

AI for RAN is the area of AI-RAN being developed the fastest. Our customers benefit from AI already today in Nokia RAN solutions and services. We use AI for RAN to reduce energy consumption, increase network capacity, reach autonomous operations level 4 and optimize user experience. We also use AI within the entire network lifecycle, including network design, deployment and optimization services.

Operators have achieved outstanding results using our MantaRay AutoPilot’s extensive AI and automation capabilities in their commercial radio access networks. For example, in 2025, MantaRay Autopilot performed 15,000 autonomous operations per hour to increase available network capacity and optimize user data rates with massive ongoing network traffic in a commercial RAN network. 

Among the hottest research topics in the AI-RAN Alliance are areas such as Massive MIMO optimization and radio channel estimation. At Mobile World Congress 2025, we demonstrated channel estimation powered by AI in our AirScale baseband. 

With advances in AI models and computing capabilities, the width and depth of AI workloads within RAN will grow over time. Nokia RAN provides the scalability needed for AI.