But according to Lee Myall, CEO of wholesale fibre provider Neos Networks, raw compute power is only part of the story. Without serious attention to the fibre networks that connect supercomputers to the rest of the economy, the benefits risk being constrained.

“The UK government’s £36 million investment to expand the Cambridge DAWN supercomputer sixfold is a welcome commitment to strengthening the UK’s AI capabilities,” Myall says.

“With access to the world’s most advanced AI processors, researchers will be able to work with bigger datasets and more ambitious projects than ever before. But unlocking the full value of that compute power will depend on the networks that connect supercomputers to data centres, businesses, and end users.”

Announced on Monday January 26, the funding will significantly boost one of the UK’s leading supercomputing centres. Based at the University of Cambridge, DAWN already plays a central role in the national AI Research Resource (AIRR), which provides free access to high-performance computing for UK researchers, start-ups and small businesses.

The expansion will increase DAWN’s capacity sixfold and introduce access to AMD Instinct MI355X GPUs, integrated by Dell Technologies and supported by a UK-developed AI software stack from StackHPC.

The government sees the move as a cornerstone of its wider AI Opportunities Action Plan, which includes more than £2 billion in public investment in compute infrastructure and a commitment to expand AIRR twentyfold by 2030. The Cambridge investment also reinforces the city’s position at the heart of the Oxford–Cambridge corridor, one of Europe’s most significant clusters for science, technology and innovation.

DAWN is already delivering tangible results. More than 350 projects have used the system, including work on AI-driven approaches to personalised cancer vaccines and research into environmental change. With additional capacity expected to come online as early as spring 2026, ministers say the enhanced system will support faster disease detection, improved public services and more accurate climate modelling.

Yet Myall argues that these ambitions will only translate into real-world impact if connectivity keeps pace with compute. “Our recent research found that 82% of data centre operators have delayed site builds or expansion due to fibre availability,” he says. “Meanwhile, 95% say access to new high-capacity fibre will directly influence where and how they invest next. Clearly, fibre has become a deciding factor for the UK’s AI future.”

This challenge is becoming more acute as data centre development shifts away from traditional metropolitan hubs. Rising land costs, power constraints and sustainability considerations are pushing new facilities into regional locations. While this decentralisation brings benefits, it also places greater demands on national fibre infrastructure.

“As new data centre sites are built outside major cities due to cost and energy considerations, key terrestrial routes linking regional hubs from Cambridge to Edinburgh should be considered essential to the success of these sites,” Myall says. “AI requires high-capacity optical and dark fibre capacity to manage HPC workloads. By investing in a robust national network, we can ensure the UK turns world-class compute into real AI impact for businesses and communities.”

The government has emphasised that the Cambridge expansion also strengthens the UK’s computing resilience by diversifying the technologies underpinning national infrastructure.

Alongside DAWN, AIRR includes Isambard-AI in Bristol, with a new national supercomputer planned for Edinburgh. Together, these assets are intended to reduce reliance on overseas hyperscalers and give domestic researchers access to cutting-edge systems.

For the University of Cambridge, the funding marks another milestone in its role as a national AI hub. The institution says the expanded DAWN system will help researchers, clinicians and innovators tackle complex challenges across healthcare, climate science and public services, while continuing close collaboration with industry partners.

From a network perspective, however, the scale and distribution of these investments underline the need for long-term planning. Supercomputers do not operate in isolation. Their value lies in how effectively they can exchange data with other research facilities, data centres, enterprises and public sector users.

The question now, as Myall frames it, is whether digital infrastructure policy will move with the same urgency.

RELATED STORIES

AI triggers job losses across the UK

New £10M Nvidia Grace-based supercomputer comes online in the UK

Autumn Budget 2025: How the UK plans to scale AI data centres