Britain is betting on AI. Now it needs the network that will run it

The UK is experiencing a watershed AI moment, and the momentum is real. British AI startups raised £6 billion in venture capital last year, and in just the first three months of 2026 they have already raised more than half that figure again.

The Government's recently launched Sovereign AI Unit, backed by a £500 million public investment, is designed to accelerate this momentum by giving the UK's most promising AI companies direct access to computing power, R&D support and government procurement opportunities.

Managing Director at Cellnex UK.

The ambitions are right. The UK has genuine strengths in AI research, a world-class university sector and a growing startup ecosystem attracting serious global capital.

But there is a fundamental question these ambitions raise, one the telecoms industry needs to put firmly on the table: you cannot scale AI without the connectivity infrastructure to support it.

The case for quality, not just coverage

Coverage maps tell you where the network is. Performance data tells you whether it actually works. More than 95% of the UK's landmass has at least some 4G coverage, more than many countries across Europe, and on that measure the UK performs well. But coverage is only part of the picture.

We all know that the UK isn’t where it should be with quality standalone 5G. The UK currently ranks last among 16 comparable high-income Western European markets on key 5G performance measures, with only 15% of 5G sites delivering the standalone architecture that is the global gold standard.

This gap has real consequences. Analysis from Analysys Mason and Opensignal shows UK users get just 5.8 Mbps of 5G speed for every euro (£0.87) they spend each month, the weakest return among comparable European markets. In Portugal, consumers get almost four times as much at 20.9 Mbps per euro (£0.87).

Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed!

But the crucial point is this: we know exactly what the problem is, we know what the solution looks like, and we have the tools to close the gap. The good news is that there is a smarter, faster and more cost-effective path to the 2030 standalone 5G target than the one the UK is currently on.

AI needs physical infrastructure, not just software

This is where the conversation needs to get specific. AI is not just software in the cloud. It requires a robust, low-latency, high-capacity network backbone to function at scale. The applications that will determine whether the UK's AI investment translates into economic growth do not run on aspiration. They run on milliseconds and bandwidth.

Delivering AI at scale requires three things working together. The first is standalone 5G architecture for ultra-low latency below 10 milliseconds and the massive bandwidth capacity that real-time applications demand.

The second is dense small cell networks, with AI applications requiring cells significantly closer to users than traditional macro towers allow. In dense urban areas, high-quality standalone 5G requires between 37 and 96 small cells per square kilometer.

The UK currently has approximately 5,000 deployed across the entire country, leaving significant room for targeted investment to close that gap. And finally, edge computing infrastructure positions processing power closer to where data is generated, enabling the real-time responsiveness that applications like autonomous vehicles and industrial automation require.

Good 5G is not about coverage percentages. It is about consistent experience, whether you are a start-up looking to scale, a commuter on a train, or a fan at a football stadium. Getting this right is what will allow the AI companies attracting record levels of capital to scale their most demanding applications on home soil.

The most efficient path to closing the investment gap already exists. Shared infrastructure is cheaper and faster to deploy, requires fewer sites, consumes less energy and, critically, it works.

Shared models have already freed up £3.4 billion for UK operators and €26 billion across Europe to reinvest in network quality. Rather than spreading capital across redundant deployments, sharing allows operators to direct investment where it matters most: performance and coverage quality.

Unlocking shared infrastructure at scale requires a planning system that supports it. The Product Security and Telecommunications Infrastructure Act was passed over three years ago and remains largely unimplemented. Full implementation would unlock upgrades on over 6,200 sites today.

Treating mobile and fixed broadband equally under Permitted Development Rights would remove an unnecessary barrier to deployment. Creating an environment which rewards quality-focused investment such as a structure that supports sharing over sole-use builds would further accelerate progress. These are practical steps; the levers are already there.

The opportunity is there to be seized

A fully realized standalone 5G network could contribute £159 billion to the UK economy by 2035. That is not a distant or theoretical prize. It is the direct return on infrastructure investment decisions being made right now.

The UK has everything it needs to lead: world-class AI talent, growing capital flows, genuine government commitment and an infrastructure sector ready to deploy shared solutions at pace.

The delivery choices made between now and 2030 will determine whether the UK's AI ambitions translate into genuine economic leadership. This country has every reason for confidence.

Fix the network, back shared infrastructure, clear the planning barriers, and the applications the UK is funding today will scale here, on infrastructure built for the job. That is an outcome well within reach.

This is no longer a technology question. It is an execution question – and one the UK is well placed to answer.

We've featured the best AI tool.

This article was produced as part of TechRadar Pro Perspectives, our channel to feature the best and brightest minds in the technology industry today.

The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/pro/perspectives-how-to-submit