Our Executive Director of Energy Transition, Griff Thomas, on what the Government’s new Data Centre Covenant gets right, and where it now needs to go further.
The publication of the Government’s new Data Centre Covenant should be welcomed.
For perhaps the first time, we are seeing a genuinely joined-up acknowledgement that data centres are not simply large commercial buildings consuming a lot of electricity. They are becoming part of Britain’s critical national infrastructure, fundamental to AI, economic growth, public services and ultimately national security.
The Government is clear about the scale of that ambition: it estimates that the UK will require 6GW of AI data centre capacity by 2030 to protect our national interests and maintain access to the compute infrastructure on which an increasingly digital economy will depend.
That is a significant statement of intent. But there is an equally significant question behind it:
Where is the power going to come from?
You cannot build a digital economy on a connection date. The biggest constraint facing many UK data centre developments is no longer capital, customer demand, or even necessarily land.
It is power.
Across the projects and developers we are working with, the same issue appears repeatedly. A site can have planning potential, fibre connectivity, investors, and a credible end user – only to discover that the electricity capacity it requires may not be available for several years.
The Covenant deserves credit for recognising this reality rather than pretending that conventional grid reinforcement alone can keep pace with AI.
The Government now proposes interventions, including reallocating and reserving grid capacity for strategically important projects, identifying capacity that could become available before 2030, and accelerating strategic connections. More significantly, it explicitly proposes a framework allowing temporary on-site generation to bridge the gap until the grid connection arrives.
That is potentially a very important change. Because the uncomfortable reality is that the global AI race is not going to wait patiently for the British electricity network.
“Bring your own power” should not be treated as an admission of failure.
There is sometimes a tendency to regard on-site generation as somehow competing with the electricity grid or undermining the transition to clean energy. That is too simplistic.
A modern data centre energy strategy could combine grid electricity, on-site generation, renewable procurement, storage, demand flexibility, fuel cells, gas generation, waste-heat utilisation, and ultimately lower-carbon gases or carbon capture. The correct solution will differ from site to site. The important thing is that we start treating the energy system surrounding a data centre as infrastructure that can be engineered, rather than assuming the only acceptable answer is to submit a grid application and wait.
Government appears to be moving in exactly that direction.
It is also proposing greater freedom for developers to self-build connection infrastructure and looking at more flexible ways for data centres to connect while maintaining system security. These are pragmatic ideas and should be developed quickly.
But here comes the challenge – The Covenant describes temporary on-site generation as something available only to “exceptionally strategic” projects, and where other interventions cannot provide capacity quickly enough. There is an obvious danger that we create a sensible solution and then surround it with so many gateways, definitions, and approvals that very few projects can actually use it.
If Britain genuinely wants 6GW of AI data centre capacity by 2030, the test should surely be whether a project can demonstrate a safe, resilient, environmentally responsible, and economically credible energy solution – not whether it is fortunate enough to pass an undefined test of exceptionalism.
And gas should not become a forbidden word. This is another area where a degree of engineering realism is required.
The transition to a low-carbon energy system is essential. But energy policy also has to recognise engineering timescales. There are locations in the UK where the gas transmission or distribution system can potentially provide very significant quantities of energy considerably sooner than equivalent electrical capacity can be reinforced.
Modern high-efficiency gas generation, reciprocating engines, turbines, and solid oxide fuel cells can therefore provide a bridge to grid power – and potentially form part of a longer-term resilient microgrid architecture. That does not mean abandoning decarbonisation. It means designing the transition.
Projects should be considering from day one how generation can progressively incorporate biomethane, hydrogen, where technically and economically viable, carbon capture, heat recovery, and increasingly renewable grid electricity.
We should be asking: “How do we make this solution progressively cleaner?”
rather than: “Can we pretend the energy requirement doesn’t exist until the grid catches up?”
The latter will not reduce global emissions. It may simply encourage investment to go somewhere else.
The Covenant is right to expect something in return. There is also an important quid pro quo in the Government’s approach. If communities are going to accommodate nationally important data centre infrastructure, they should benefit from it. The Covenant expects strategic developments to produce Local Benefit Plans, with investment linked to the scale of development and communities having a meaningful role in deciding how those benefits are used. That is sensible.
Indeed, the UK can probably go further.
Major US infrastructure developments increasingly use structured community-benefit arrangements that link investment to tangible local outcomes. There is no reason why large UK data centre campuses cannot make similar long-term commitments around skills, apprenticeships, local supply chains, community infrastructure, and employment. But those commitments need to produce genuine benefits rather than simply becoming another planning levy.
The Covenant also rightly expects developers to invest in paid apprenticeships, colleges and universities, recruit locally where possible, and strengthen UK supply chains. For an infrastructure industry facing significant skills shortages, that could be one of the most valuable aspects of the policy.
Environmental standards need engineering behind them too. The Covenant proposes strong expectations around water consumption, energy efficiency, emissions reporting, and waste heat. Again, the direction is right. New facilities are expected to target a PUE of 1.3 or better, minimise dependence on public water supplies, and investigate productive uses for waste heat. But waste heat is a good example of why policy must remain grounded in engineering.
Writing “reuse the heat” into a strategy is easy. Finding a viable heat customer, at the right temperature, sufficiently close to the data centre, with a matching demand profile and an economic heat network is rather more complicated.
The answer is not to abandon the ambition. It is to bring infrastructure engineers into these developments early enough that gas, power, water, heat, planning, and environmental constraints are designed together rather than considered sequentially.
The real opportunity is bigger than the data centre. Done properly, this could become a significant UK industrial opportunity.
These projects need substations, electrical networks, gas infrastructure, pressure reduction stations, generation equipment, control systems, civil engineering, fibre, water infrastructure, heat networks, environmental engineering, and highly skilled people to design, build, operate, and maintain them.
That means British engineering jobs. It means apprenticeships. It means opportunities for UK manufacturers and technology businesses. And it means infrastructure investment in parts of the country that have not always benefited equally from previous waves of digital growth.
The Covenant’s emphasis on communities, UK supply chains, skills, and reindustrialisation is therefore every bit as important as its provisions on data centres themselves.
Now we need delivery to move at the speed of technology. The Government should be applauded for recognising the problem.
The Covenant represents a noticeable shift from asking simply “how do we connect more data centres?” towards the much more useful question: “How do we create the energy infrastructure that allows strategically important data centres to be built?”
That is progress. But AI infrastructure is developing at an extraordinary speed. Capital will move internationally. Equipment manufacturing slots will be taken. Developers will choose countries where there is greater certainty over power and programme.
So the next phase matters enormously.
We now need rapid clarity around what constitutes a strategic data centre, how temporary generation will be assessed, how environmental permitting will interact with accelerated planning, how self-build network infrastructure will operate, and – critically – how quickly decisions will actually be made. Because there would be a certain irony in publishing a policy designed to accelerate data centres and then spending two years developing the process for accelerating them.
Britain has the engineering capability, energy infrastructure, technology companies, and international investment appetite to become one of the world’s leading locations for AI and data centre infrastructure.
At United Infrastructure, we are already working with developers, technology providers and major data centre customers on exactly these challenges, from grid and gas connections to on-site generation, microgrids, electrical infrastructure and the move to lower-carbon energy.