There is no doubt the UK needs more computing infrastructure, but there remains uncertainty over the type of infrastructure necessary to deliver compute across the country.
Currently, most of the debate focuses on where datacentres should be built and how quickly they can be connected to the grid.
While these are important questions, it’s equally important not to skip over a more fundamental one: whether we are planning for the right future at all.
In working with energy networks, technology companies and policymakers on some of the UK’s most pressing infrastructure challenges it has struck me how differently we think about energy infrastructure and digital infrastructure.
In the energy sector for example, we’ve learned the answer isn’t always to build more. We’re increasingly considering whether we can make better use of existing assets through flexibility, optimisation and better visibility of performance. Yet when it comes to datacentres, the conversation often jumps straight to capacity and grid connections, which feels like a missed opportunity.
Rather than viewing datacentres simply as large electricity consumers, we must start thinking about them as part of a wider energy, digital and industrial system, and starting with evidence over assumption.
Start with evidence, not assumptions
The UK had around 1.6GW of operational datacentre capacity in 2024, with government ambitions pointing towards at least 6GW of AI-ready capacity by 2030. At the same time, reports suggest around 50GW of prospective datacentre demand is sitting in electricity connection queues, highlighting the growing gap between ambition and deliverability.
So, before deciding how many datacentres the UK needs, we need a much clearer view of actual compute demand. Forecasts vary enormously, planning data sits across hundreds of local authority portals, and grid connection queues contain lots of projects that may never be built. It’s surprisingly difficult to get a complete picture, and the uncertainty is reflected globally.
The International Energy Agency estimates datacentre electricity consumption could rise from around 415TWh in 2024 to almost 945TWh by 2030, largely driven by AI.
The UK needs a shared evidence base that brings together demand forecasts, planning applications, grid connections and energy infrastructure data. Otherwise, we risk planning for headlines rather than reality and viewing digital and energy infrastructure in siloes. This is where convening capabilities across the industry will prove critical to establishing a shared framework that can bring together experts across the energy ecosystem, and unlock opportunity for meaningful collaboration and strategic partnership.
Plan digital and energy infrastructure together
Although datacentres are increasingly critical infrastructure, they’re often planned separately from the energy systems that support them. Future siting decisions should consider renewable generation, network constraints, water availability and local economic opportunity alongside land availability.
Ireland shows why this is important, and serves as a useful warning. In 2023, datacentres accounted for 21% of the country’s total metered electricity consumption, a huge jump from just 5% in 2015. While the UK may not necessarily follow the same path, Ireland’s experience demonstrates what can happen when digital and energy planning become disconnected. In the UK we have an opportunity to take a more integrated approach from the outset, which is why we must think of datacentres as part of a broader digital and industrial system.
Treating compute as a flexibility asset is where the biggest opportunity lies for the UK. In delivering pioneering innovation and acceleration programmes with energy networks to enable deep tech companies to scale, I’ve seen first-hand how digital technologies are changing our understanding of flexibility. Batteries, EV charging infrastructure and industrial processes are increasingly expected to respond dynamically to system conditions. Compute infrastructure is one of the few major electricity loads that is rarely discussed in those terms. Yet some compute workloads are far more flexible than many people realise.
Google for example, has already demonstrated carbon-aware computing systems that can optimise workloads by location and time based on the availability of low-carbon electricity. Tasks such as media processing for YouTube and Google Photos can be shifted without users noticing any difference. Instead of always asking how the grid can accommodate more compute demand, perhaps we should also ask how compute demand can help support the grid.
Look beyond hyperscale; build for resilience and innovation
Hyperscale campuses will remain an important part of the picture, but I’m not convinced they’ll be the whole story. In conversations with network operators, renewable-energy developers and technology providers, there is growing interest in whether future compute infrastructure could be distributed across facilities that operate at different scales. This is particularly relevant in renewable-rich regions where grid constraints often limit the ability to make full use of low-carbon generation. Modular facilities, edge computing and distributed architectures could help reduce pressure on transmission networks, improve resilience and make better use of renewable energy that might otherwise be curtailed. The future may look less like a handful of giant facilities and more like a connected ecosystem of compute infrastructure.
The future of computing won’t be shaped by AI alone. New processor architectures, cyber threats, supply-chain dependencies and emerging technologies such as quantum computing could all change today’s assumptions. Rather than trying to predict a single future, the smarter approach is to build infrastructure that can adapt and evolve. That’s why the UK should support real-world demonstrators that explore flexible computing, modular datacentres, heat reuse and closer integration between digital and energy infrastructure. Innovation works best when ideas can be tested in practice rather than debated in theory. Digital Catapult, for example, provides partners with access to testbeds, facilities and expertise to validate new solutions in a safe and secure environment.
The UK’s datacentre challenge is often presented as an electricity-demand problem. That’s too narrow a framing. The biggest opportunities often emerge when two sectors start solving problems together. Too often, the future of computing and the future of energy are treated as separate conversations. In reality, they are becoming increasingly interconnected.
As global datacentre electricity demand heads towards 945TWh by 2030, countries that simply build more infrastructure will benefit. But, countries that rethink the relationship between compute, energy and industrial strategy could create a much more durable advantage. The goal shouldn’t be to build the most datacentres, but to build the most intelligent, flexible and resilient digital infrastructure system in the world.
Luke Sperrin is head of clean energy industries at Digital Catapult.
S 004