New research reveals a growing tension at the heart of the UK’s AI ambitions: businesses are confident they can scale AI workloads, but network infrastructure is already shaping where and how that growth can happen.

New research found that 78% of large enterprises across financial services, manufacturing, life sciences, media and energy, believe their current connectivity is mostly or fully prepared to support a 2–3x increase in AI workloads over the next year. However, just 3% say regional network infrastructure has had no impact on their AI plans. This includes private AI models, corporate-data applications, training environments and inference services that require low latency and high performance.

The impact is already translating into real-world compromises: 38% of companies have kept workloads more centralised than originally planned, while 34% have moved workloads to another UK region. Others have been force to either change their infrastructure provider (37%), while some (36%) have used cloud instead of regional or edge deployment. About a third (32%) have reduced deployment scale, while 29% have delayed deployment as a result of poor network infrastrcuture.

AI opportunity is becoming increasingly regional

Regional infrastructure deployment is becoming more significant as the UK’s AI opportunity spreads beyond London. While the capital remains the strongest centre of gravity, identified by 42% of enterprises as relevant to their AI strategy over the next two to three years, the North West (27%), Midlands (26%), East of England including Cambridge (24%) and Scotland (24%) also feature strongly.

Connectivity will increasingly determine whether that regional opportunity can be realised.

Ninety-two percent of enterprises say local connectivity availability influences where they deploy or expand AI workloads, while nearly half (47%) say AI workloads typically need to be deployed within the same UK region as end users to achieve the required latency and performance. A further 22% need them within the same metro area. Those pressures are set to increase as AI moves closer to users, applications and operational environments.

Almost three-quarters (74%) of enterprises expect a shift towards inferencing over the next two to three years, while every enterprise surveyed expects AI to increase its network capacity.

Lee Myall, CEO of Neos Networks, said: “AI is perhaps the most important technology of our lifetimes, and the UK has every reason to be confident about its AI opportunity, but confidence and readiness aren’t the same thing.

Recommended reading

• Ofcom Launches Mobile Coverage Checker
• UK Economy Missing Out on £8.6BN Due to Connectivity Gaps
• Digital Connectivity Vital Part of Scot Gov’s National Islands Plan
• New UK Deal to Eliminate Mobile Blackspots on Busy Rail Routes

“Our research shows businesses believe they can accommodate significant growth, yet almost all are already adapting their plans around the infrastructure available to them. As AI becomes more distributed and more regional, those compromises will become increasingly difficult to ignore.

“This Government has talked about delivering good growth in every postcode. If we’re serious about that ambition, we need a much more joined-up view of the infrastructure that makes growth possible. Power and compute are of course essential, but so is the high-capacity connectivity that lets data and workloads flow between data centres, businesses and consumers.

“Without that foundation, we risk repeating a familiar story across the UK; one of regional divides and disparities. We need to avoid a scenario where the places with the strongest infrastructure capture the AI opportunity and others are forced to centralise, relocate or delay. The goal has to be giving every region the infrastructure to participate in the AI economy, rather than allowing the network to decide where growth can happen.”