Cloud AI-RAN

The world is in the early years of an AI supercycle, a multi-decade transformation where AI becomes pervasive, driving exponential demand for intelligence, advanced connectivity and next-generation networks. 

Current telecommunication providers’ networks are not designed to meet the demands and scale of AI traffic. Our Cloud AI-RAN is engineered to be AI-native from the ground up, providing a clear evolution path to 6G.

Our Cloud AI-RAN is built on the flexible anyRAN software, which is open, cloud-native and hardware-agnostic. The anyRAN software runs across GPU, ARM and x86 architectures, supporting open interfaces and multi-vendor ecosystems and it is available with our AirScale and Cloud RAN solutions. This ecosystem flexibility accelerates innovation and protects our customers’ investments.

In our reference Cloud AI-RAN configuration we have GPU which is used to run the anyRAN baseband SW together with CPU. This GPU accelerated platform can be hosted on industry-standard COTS servers.  In this solution, the hardware layer is abstracted, enabling telecommunication providers to benefit from feature consistency across Cloud AI-RAN and AirScale AI-RAN.

This enables scalable performance for both AI-native RAN workloads and broader ecosystem of AI applications while maintaining open standards and reference architectures for partner and third-party integration.

We are driving the evolution toward AI-native networks with proven software innovation, strategic partnerships and industry-leading research into AI-ready infrastructure. With our Cloud RAN projects and proofs of concept, we are leading the evolution toward the Cloud AI-RAN era.

We are working across standards bodies and open ecosystem initiatives, and co-innovating with telecommunication providers to ensure they can capture the full value of the AI supercycle, both technically and commercially.

Cloud AI-RAN turns the network into a platform for innovation, going beyond connectivity and enabling new monetization opportunities for telecommunication providers. They can monetize AI-related traffic optimization and near-real-time responses, offer differentiated service tiers, and use excess GPU capacity for thereby optimizing TCO.

• AI for RAN: Bring operational and spectral efficiency to the next level through massive AI-driven optimization, increasing your network performance in the AI-native RAN era.
• AI on RAN: Monetize agentic AI frameworks and near-real-time RAN exposure via APIs through running AI apps for industries with strict latency and data locality requirements.
• AI and RAN: Monetize on-demand GPU-as-a-Service solutions with spare GPU capacity at RAN Edge to host AI applications and maximize ROI (Return on Investment).

Mobile operators use Cloud RAN in their commercial networks. Nokia Cloud RAN is field-proven in close collaboration with customers and technology partners. Read our case studies and papers to understand how mobile operators have trialed and implemented Cloud RAN.