AI steps onto the live 5G stage in the UK
In a first for the country, O2 has teamed up with Nokia to run an artificial‑intelligence‑enhanced radio access network (RAN) on its Giga Site at London Paddington. The site, one of the busiest transport interchanges in the capital, now hosts a system that can predict traffic spikes and re‑allocate radio resources without human input. The trial, which ran on O2’s commercial 5G network, is being billed as a practical demonstration that AI can squeeze more capacity out of the same spectrum.
How the AI‑RAN works
Traditional RANs rely on static configuration tables that are tuned by engineers on a weekly or monthly basis. Nokia’s solution replaces those static rules with a machine‑learning model that continuously ingests performance metrics – such as signal‑to‑noise ratio, user load, and interference levels – and forecasts short‑term demand. When the model anticipates a surge (for example, a wave of commuters boarding a train), it automatically adjusts parameters like carrier aggregation and beamforming to maximise the use of available spectrum.
The core of the system is a predictive AI engine embedded directly in the radio hardware. Rather than sending data to a distant cloud for processing, the AI runs at the edge, keeping reaction times measured in milliseconds. This edge‑centric approach is crucial for mobile networks, where latency and reliability are paramount.
Measurable gains from the live test
During the eight‑day trial, O2 reported a spectral efficiency uplift of roughly 20 %. In plain English, the site was able to move about one‑fifth more data through the same radio channels, effectively delivering higher download speeds and smoother video streaming without needing extra spectrum licences or new antennae.
Key performance highlights included:
- Throughput spikes of up to 30 % during the busiest hour of the morning rush.
- Stable latency even when the number of active devices doubled.
- Reduced buffering for video services, a benefit that will be noticeable to commuters watching on‑the‑go.
These figures were captured on a live consumer network, meaning ordinary O2 customers at Paddington experienced the improvements directly.
Why this matters for Android phone users and contract shoppers
For the typical Android owner in the UK, the headline is simple: faster, more reliable 5G when you need it most. The AI‑RAN is designed to shine in high‑density environments – train stations, stadiums, city centres – where traditional networks can become congested. If O2 rolls the technology out across its wider 5G footprint, users can expect:
- Higher download speeds for large app updates and game downloads.
- Smoother streaming of services like YouTube, Netflix, and TikTok, even during peak travel times.
- Lower latency for real‑time gaming and AR/VR experiences that are becoming common on newer Android devices.
- Potentially better contract value, as O2 can offer higher‑performance data plans without having to invest in additional spectrum.
For those hunting the best deals, the rollout could translate into more competitive data allowances or lower monthly fees, because the network can serve more traffic with the same resources.
O2’s broader £700 m Mobile Transformation Plan
The AI trial is just one piece of O2’s £700 million Mobile Transformation Programme announced earlier this year. The plan combines capital spending on new sites, spectrum refarming, and smarter software to lift overall network reliability and capacity. By embedding AI at the radio layer, O2 hopes to future‑proof its infrastructure against the inevitable rise in mobile data consumption – a trend driven by 5G‑only smartphones, cloud gaming, and the rollout of 8K video.
Jeanie York, O2’s Chief Technology Officer, summed up the ambition: “Mobile data demand continues to grow, so we need smarter ways to make the most of the spectrum and infrastructure we already have.” The AI‑RAN trial, she added, shows that intelligent automation can deliver tangible performance gains while keeping operational costs in check.
Nokia’s perspective and the industry outlook
Emma Falck, President of Mobile Networks at Nokia, positioned the test as a milestone in moving AI from the lab to live networks. She noted that embedding intelligence directly into the RAN “creates a stronger foundation for future AI‑driven services,” hinting at possibilities such as automated fault detection, energy‑saving modes, and even network‑wide orchestration of edge‑computing resources.
Industry analysts see this as part of a broader shift. As 5G matures, operators are looking beyond raw spectrum and towards software‑defined optimisation to stay competitive. AI‑enabled RANs could become a standard feature in the next generation of network upgrades, especially as the UK government pushes for nationwide 5G coverage by 2027.
What’s next?
O2 has not disclosed a timeline for a wider rollout, but the company indicated that the Paddington trial will inform a phased deployment across other high‑traffic sites. If the performance gains are replicated elsewhere, the technology could become a selling point for O2’s upcoming contract bundles, particularly those targeting heavy data users and early‑adopter Android enthusiasts.
For now, commuters passing through London Paddington may already be noticing a smoother 5G experience. As the AI‑RAN moves from pilot to production, the benefits are likely to ripple out to the broader O2 network, offering UK Android users a more responsive and future‑ready mobile experience.
All statements from O2 and Nokia are taken from their joint press release dated 26 October 2026.
