AI-RAN Puts Network Functions and AI Services on Shared Infrastructure
Nokia and NVIDIA argue that telecom operators can turn network infrastructure into a shared platform for radio access and AI workloads, creating a path from programmable 5G toward 6G. Their AI-RAN collaboration pairs Nokia software with NVIDIA accelerated computing; the companies say it can improve network performance, support AI services at the network edge and help automate network operations.
One shared platform is meant to carry both network and AI workloads
Nokia and NVIDIA’s proposal is to run radio access network functions and AI workloads on shared accelerated computing infrastructure. The aim is not only to improve 5G, but to make the network software-defined and programmable enough to evolve toward 6G.
The architecture shown pairs NVIDIA’s Aerial RAN Computer with Nokia’s CUDA-accelerated RAN software. Its labeled layers place 5G and 6G alongside internal AI workloads and AI services. Nokia and NVIDIA describe the system as optimized for signal processing and AI inference: the same infrastructure is presented as supporting radio operations and services that use AI. Operators could use network infrastructure not only to connect devices, but also to run AI applications and offer services built on them.
Nokia and NVIDIA say the collaboration has moved from vision to deployment within a year: Nokia anyRAN software is running on NVIDIA’s accelerated computing platform. A chart presents spectral efficiency as improving through Nokia AI-RAN subscriptions, from a baseline of 1× to a projected 1.5× in 2027 and 2× in 2028. The figures are framed as a forward-looking benefit of the subscriptions, alongside the broader claim that AI-RAN can improve network performance.
With AI inference running at the network edge, telecom operators could provide AI services as well as connectivity. Nokia and NVIDIA present this as a commercial opportunity alongside the projected radio-performance gain: a shared platform for network operations and AI services.
One platform, software-defined, programmable, built for the AI and 6G future.
The service examples range from voice assistants to industrial robotics
Nokia and NVIDIA present demonstrations across voice and vision applications, industrial robotics, healthcare access and emergency response. These are examples of services the partners say AI-RAN could support; they are not, on their own, evidence that each service is deployed at scale.
Elisa’s Memory Lightener demonstration shows a user speaking a request to book a dentist appointment while the application displays a live transcription. Nokia and NVIDIA describe the use case as bringing natural conversation to AI assistants. In a separate T-Mobile and Linker Vision demonstration, a dashboard tracks city traffic and flags a stalled vehicle, with an AI-generated analysis of road and traffic conditions. The partners describe vision AI as a way to help cities become safer.
The emergency-services demonstration combines a caller’s speech with an AI-generated summary for a call-center operator. The displayed summary concerns a request for safety guidance during a wildfire and advice about a possible evacuation. Nokia and NVIDIA characterize the application as real-time voice AI for emergency services. Indosat Ooredoo Hutchison is associated with expanding access to healthcare in remote areas; the accompanying footage shows people in a waiting room before transitioning to a map with cell towers.
The robotics examples show two different tasks. In a Deutsche Telekom and Agile Robots scene, robotic arms manipulate an electronic component, under the partners’ claim that the platform can power physical AI at industrial scale. A SoftBank and Yaskawa demonstration shows robotic arms sorting objects, described as coordinating autonomous robots in real time. Together, the examples illustrate the range of applications Nokia and NVIDIA associate with AI-RAN, from language and vision systems to machines acting in physical environments.
AI agents are also meant to help operate the network
Nokia and NVIDIA frame AI as a tool for running the network itself, not only for providing services over it. A Nokia autonomous-networks dashboard depicts network-health and service-quality indicators during a tennis event, alongside an assistant identifying a likely incident root cause across radio and core systems. The interface also shows a proposed service intent to maintain priority voice and data service for VIP users. The action is marked autonomous and human-in-the-loop, with medium risk.
The dashboard connects diagnosis to operational decisions: the assistant identifies a possible cause, displays service impact and a proposed priority, and indicates that human oversight remains part of the process. Nokia and NVIDIA say AI agents can help operators keep networks reliable and secure. In the demonstrated scenario, the agent supports incident analysis and service prioritization as network conditions change.
Nokia and NVIDIA name T-Mobile, BT, Vodafone, Elisa, du, stc, SoftBank, docomo, Indosat and TPG as operators collaborating to bring AI-RAN into 5G networks. Their stated ambition is a network foundation that is programmable for 5G today and ready for 6G.