The UK’s ambition to become a global leader in digital infrastructure is well founded. Artificial intelligence, cloud computing, scientific research, financial services, public sector digitisation and advanced manufacturing all depend on resilient, high-performance datacentres.
The question is no longer whether we need more capacity. We undoubtedly do. The real question is whether we are planning for that future using sound evidence rather than speculation.
At present, the answer is probably no.
Debate dominated by assumptions
Much of the current debate has become dominated by headline figures. Government ambitions for gigawatts of datacentre capacity, forecasts of electricity demand, projections around water consumption and competing estimates of future AI growth all attract attention. Yet many of these numbers represent aspirations, worst-case scenarios or early-stage requests rather than realistic delivery plans.
That creates a significant risk that infrastructure decisions are made using assumptions that may never materialise.
Take electricity demand. Ofgem has reported that around 50GW of prospective datacentre demand is currently sitting within the electricity grid connection pipeline.
On paper, that appears enormous. In practice, anyone familiar with infrastructure development knows that not every project requesting a grid connection will ultimately be built. Some schemes will be redesigned, delayed or cancelled altogether. Others may secure planning permission but never obtain funding. Some will be overtaken by technological developments that reduce their power requirements before construction even begins.
Treating every application as inevitable risks overestimating demand and misallocating scarce investment.
Fragmented planning
Planning applications present a similar challenge. Across the UK, information is fragmented across hundreds of individual local authority planning portals. There is no comprehensive national picture showing which datacentre developments have been proposed, approved, under construction or completed. Even when applications can be identified, determining which projects will genuinely reach operation is far from straightforward.
Without reliable visibility of the development pipeline, forecasting future capacity becomes increasingly difficult.
Water illustrates another area where careful interpretation matters. Datacentres are frequently criticised for their potential water consumption, particularly where evaporative cooling is used. Water resilience is undoubtedly an important planning consideration, especially as climate pressures increase.
However, discussions often rely on headline figures that represent extreme scenarios rather than likely outcomes. Modern datacentre design increasingly incorporates closed-loop cooling systems, air cooling, liquid cooling technologies and sophisticated water efficiency measures. Future facilities are unlikely to resemble those built even five years ago. Rational planning requires understanding the technologies actually being deployed, not assuming every future facility will use yesterday’s engineering approaches.
How to plan for uncertain scenarios
The pace of technological change presents perhaps the greatest planning challenge of all.
Artificial intelligence has undoubtedly accelerated demand for compute capacity. At the same time, semiconductor manufacturers continue to deliver substantial improvements in performance per watt, while server densities, cooling technologies and workload optimisation continue to evolve rapidly. The amount of computing delivered from a single rack continues to increase dramatically.
Forecasting infrastructure requirements 10 or 15 years ahead therefore becomes exceptionally difficult. We know demand will grow, but predicting exactly how much physical infrastructure will be required becomes increasingly uncertain as technology itself changes.
That uncertainty should not lead to paralysis. Instead, it argues for more adaptive planning.
Evidence-led planning required
Rather than attempting to predict a single “correct” future, government should adopt a dynamic planning framework that is continually updated using real-world evidence. Grid capacity, planning approvals, construction progress, occupancy rates, technological improvements and regional demand should all feed into an evolving national picture.
This is common practice in many other infrastructure sectors.
Transport planners continuously revise passenger forecasts. Energy planners regularly update generation assumptions. Water companies adapt investment plans as consumption patterns change. Datacentre planning deserves the same evidence-led approach.
Greater transparency would significantly improve decision making. A national database of datacentre planning applications, project status and operational capacity would provide policymakers, network operators and investors with a far clearer understanding of where genuine demand exists. Standardised reporting of energy efficiency, cooling technologies and expected utility requirements would further improve forecasting accuracy.
Equally important is recognising that datacentres should not be viewed in isolation.
Their location influences electricity networks, water infrastructure, fibre connectivity, transport, skills availability and local economic development. Strategic planning therefore requires coordination across multiple government departments, regulators, utilities and local authorities. The same joined-up thinking is essential when facilities reach the end of their operational life, where secure datacentre decommissioning ensures infrastructure is retired responsibly, sensitive data is protected and valuable equipment can be reused or recovered wherever appropriate. Decisions made independently often create unintended bottlenecks elsewhere in the system.
Sophisticated technology needs sophisticated planning
The UK has an opportunity to become one of Europe’s most attractive locations for digital infrastructure. We have a strong technology sector, world-class connectivity, respected regulatory institutions and significant expertise across engineering, finance and research. But maintaining that position requires planning frameworks that are as sophisticated as the technologies they seek to support.
Ultimately, rational planning is not about selecting the highest demand forecast or the lowest environmental estimate. It is about recognising uncertainty, improving data quality and making decisions that can adapt as circumstances evolve.
Digital infrastructure has become as fundamental to the economy as roads, railways and electricity networks. We would not plan those systems using incomplete information or worst-case assumptions alone. Nor should we assume that every proposed project, every grid connection request or every demand forecast will become reality.
Instead, the UK needs a planning model built on evidence, transparency and continual refinement. Better national data, improved coordination between planners, utilities and industry, and regular reassessment of infrastructure requirements will produce better outcomes than static forecasts ever can.
The future will almost certainly require more datacentre capacity than we have today. The challenge is ensuring we build the right capacity, in the right places, at the right time, supported by the right infrastructure and informed by the right data. If we can achieve that, the UK will not simply be planning for the datacentres of tomorrow – it will be planning for a more resilient digital economy.
Daniel Smith is CEO of Astralis Technology.
