VodafoneThree joins Nokia/Nvidia test squad as carriers kick tires on AI-RAN

VodafoneThree will test Nokia and Nvidia’s AI-RAN platform, initially in the lab before field trials, to assess radio performance gains and whether GPU-powered infrastructure can support network slicing and third-party AI workloads.
In sum – what to know:
VodafoneThree trial – of Nokia’s anyRAN software on Nvidia’s Aerial platform, testing potential improvements in capacity, efficiency, alongside new apps for slicing and edge AI.
Expanded roster – of triallists for Nokia’s AI-RAN bundle includes O2, BT/EE, T-Mobile, and Orange, as telcos move from laboratory experiments towards live-network deployments.
Commercial viability – remains an open question, with operators weighing GPU power demands and costs against AI-driven improvements to conventional RAN equipment.
VodafoneThree is the latest UK operator to kick the tires on Nokia’s AI-RAN bundle, which mixes-in accelerated AI chips for smarter RAN signal processing. The firm said it will test Nokia’s anyRAN software on Nvidia’s Aerial RAN platform in a controlled lab environment initially, before progressing to field trials at selected RAN sites. It is looking to gauge uplifts in network capacity, reliability, and spectrum efficiency. The work will also explore how a GPU-based AI-RAN set-up might be leveraged for slicing, and also as a compute engine for third-party AI workloads.
It follows a bunch of parallel telco experiments with Nokia’s high-stakes Nvidia strategy. O2 said at the end of last month it had completed the UK’s “first live deployment” of AI-RAN at a site in London, claiming a 20 percent jump in spectral efficiency. Nvidia had paraded a series of carrier wins the week before, at GTC in Berlin – in Jensen Huang’s last public outing as Nvidia chief, before his resignation. These were mostly via its Nokia partnership, and included a new AI-RAN test project with UK-based BT/EE, starting next year (2027).
The pair have plenty of other exploratory deals outside of the UK, including with T-Mobile in the US; NTT Docomo and Softbank in Japan; Indosat Ooredoo Hutchison in Indonesia; Du and Etisilat in the UAE; Mobily, STC, and Zain in Saudi Arabia; A1 Group in Austria; Chunghwa Telecom in Taiwan; and TPG Telecom in Australia. Besides, T-Mobile has just made the first live AI-RAN call in the US, and Deutsche Telekom is piloting AI-RAN at distributed sites in Germany (with uplink gains of 10-15 percent in lab trials), expanding to field trials in 2027.
Elisa has run simultaneous voice, data, and video workloads on AI-RAN servers on its live network in Finland. France-based Orange, meanwhile, is testing AI-RAN for spectral efficiency and edge services, and is preparing to take Nokia’s AI-RAN platform out of the lab and into its live network, it told Intelligent RAN Forum in September. It wants to test real-world gains in spectral efficiency and energy consumption, and also explore whether the platform has legs as an engine for parallel AI workloads, beyond just RAN optimisation.
Orange will deploy all three of Nokia’s AI-RAN set-ups, it said – likely starting with plug-ins for its existing AirScale base, expanding to new AI-RAN nodes, and a cloud-native deployment. Nokia reckons its AI-RAN platform has already shown over 20 percent spectral efficiency gains via “AI-driven radio innovations”, and will deliver 50 percent gains by full launch some time in 2027, and over 100 percent by 2028, it said. The latter measure would mean telcos more-than-doubling the capacity of their existing spectrum assets. Hence the interest, of course.
But for perspective: Optus in Australia claims spectral efficiency gains in its live network, already, of up to 25 percent – at existing RAN sites, with AI tweaks from Ericsson to link adaptation, coverage prediction, and coverage compensation. As such, even positive results from all the AI-RAN trials with Nokia will be weighed by operators, ultimately, against AI-based improvements in standard RAN systems, including from rival vendors – based on custom chips and CPUs. They will also consider perceived higher-power demands from GPU-based RAN platforms.
Equally, there is a calculation to be made about putting GPUs in distributed RAN infrastructure in telco networks, as a model to support new slicing services and to host (and monetize) third-party inference workloads. VodafoneThree references both, and makes them symbiotic. Its so-called SuperMobile priority slicing service, for consumers and enterprises, offers “four-times faster” (up to) 5G speeds than standard plans “in busy areas”, and will potentially open up the network to new kinds of edge AI cases, it suggests.
It all sounds a little loose, and hopeful; but VodafoneThree stated: “As AI becomes increasingly embedded in daily life, mobile networks will need to support growing volumes of AI-driven traffic alongside traditional voice, data, and video. The programme will explore how Nvidia accelerated computing can support AI and RAN workloads on common infrastructure, helping VodafoneThree understand how network resources could be used both to enhance connectivity and support new AI-powered services closer to customers.”
Initial trial use cases include machine learning-based channel estimation and link adaptation, it said, to assess changing radio conditions and adapt how data is transmitted in real time. It is also interested to investigate: “radio optimisation, mobility, energy efficiency, anomaly detection, coverage and capacity optimisation, quality-of-service (QoS) assurance, and the sharing of accelerated compute between RAN and AI workloads”, it said. It will benchmark results against existing network performance, where the UK is falling behind.
Iain Milligan, network development and infrastructure director at VodafoneThree, said: “We’ll test these capabilities to understand how AI can improve network performance and, ultimately, the experience we deliver for our customers. As our 5G Standalone network evolves, we see significant potential for AI-RAN to help us respond more dynamically to different customer needs and unlock new, differentiated connectivity experiences, while laying the foundations for the next generation of intelligent, AI-native networks.”
Emma Falck, president of mobile infrastructure at Nokia said: “This collaboration with VodafoneThree is an important step in moving AI-RAN from industry ambition toward commercial reality. By combining Nokia’s AI-RAN innovations with NVIDIA accelerated computing, we can help VodafoneThree enhance radio performance and network efficiency today while building a software-driven foundation for new edge-AI services and AI-native 6G. It is a practical approach that protects existing network investments and creates room for continuous innovation.”
Comments are closed.