Tim Flagg
Chief Executive, UKAIThe UK does not need the world’s largest data centres – it needs the most intelligent ones. In this article, Chief Executive of the UK’s AI trade Association, Tim Flagg sets out how integrated infrastructure, smarter energy pricing and transparent standards can turn climate constraints into competitive advantage.
The recent debate around AI data centres and climate change raises an important question: can the UK expand its AI capability without undermining its environmental commitments?
The honest answer is this – it depends entirely on how we design the system.
Artificial intelligence (AI) is often discussed as if it were intangible. In reality, AI runs on physical infrastructure. It requires electricity, cooling, fibre networks, land, engineers, and long-term investment. That reality has led some to conclude that AI data centres are inherently at odds with climate policy.
That conclusion is premature.
The UK does not face a choice between climate leadership and AI leadership. But it does face a choice between fragmented infrastructure delivery and intelligent system design.
“The true risk is not that AI uses energy. The risk is that we continue to plan energy, digital infrastructure, grid access, water use and skills in silos.”
The true risk is not that AI uses energy. The risk is that we continue to plan energy, digital infrastructure, grid access, water use and skills in silos. When infrastructure is built in isolation, bottlenecks emerge, costs rise and public trust erodes. When it is designed as a connected system, constraints become drivers of innovation rather than brakes on growth.
From scale to system design
Britain’s structural conditions actually point towards a distinctive opportunity. We have high and volatile electricity prices, grid constraints, and planning sensitivities. These are often framed as disadvantages. In fact, they push us towards efficiency.
If the UK tried to compete on sheer hyperscale compute volume alone, we would struggle against countries with cheaper power. But if we compete on efficiency – AI that delivers measurable value per unit of energy consumed – we can build a globally relevant model.
This is what we mean by ‘Green AI’, as we set out in our recent report: AI, Energy and Infrastructure: Building a Green AI Superpower. AI that is energy-efficient, affordable, deployable at scale and aligned with long-term economic and environmental sustainability. Innovation drives this efficiency, across areas such as chip design, grid management, smaller language models, modular data centres and heat re-use. The UK can lead by bringing these innovations together.
The climate debate rightly focuses on electricity demand. Data centres require continuous power, and AI workloads are growing. But the UK’s problem is not simply generation capacity. It is market design.
Energy markets, not energy shortages
Wholesale electricity prices remain tied to gas through marginal pricing, even as the share of low-carbon generation increases. This means electricity prices can remain elevated and volatile even when renewables are abundant. For long-lived AI infrastructure, price volatility is as damaging as scarcity. Investors require predictable operating costs over 10 to 20 years.
Targeted reform – expanding long-term fixed-price arrangements, reducing exposure to gas-linked volatility, and investing in grid reinforcement, storage, and flexibility – would materially reduce system risk. These are not AI-specific subsidies. They are structural energy reforms that benefit households and industry alike.
Great British Energy, if used intelligently, could play a stabilising role. Not by offering preferential discounts to data centres, but by accelerating low-cost clean generation, supporting long-term contracting, and strengthening system resilience. That reduces risk for everyone.
Designing for legitimacy: water, planning and place
Water usage is another concern often raised. It is important to treat it seriously. In parts of the East and South East, water stress is real. But design choices matter. Closed-loop and dry cooling systems are already widely available. Transparent reporting on water usage effectiveness can replace speculation with evidence. Early integration of cooling strategy into planning decisions reduces the risk of local opposition driven by mistrust.
Planning reform will be critical. Speed alone is not the objective. Integration is. A data centre should not be treated as a warehouse with an unusually large plug. It should be assessed as part of a combined digital and energy system underpinning economic productivity and public services. Where projects bring together behind-the-meter renewables, storage, grid reinforcement, and connectivity as a coherent scheme, planning decisions become more robust and less contentious.
Equally important is demand discipline. Not all AI workloads deliver equal economic value. The UK should anchor infrastructure growth to high-impact use cases – health services, energy optimisation, public administration, and advanced manufacturing – rather than speculative capacity built on headline projections. Evidence-based demand modelling and aggregated public sector procurement can reduce overbuild while strengthening national capability.
Efficiency is not a constraint on ambition. It is a competitive strategy.
Efficiency as the UK’s competitive advantage
In a world where electricity, grid capacity and public legitimacy are increasingly scarce, the ability to deliver useful AI outcomes with less energy, less water and lower systemic risk will define long-term leadership. Many countries face similar, increasing structural pressures. Few can afford unlimited hyperscale abundance. The UK can become an exporter of efficient, well-governed AI systems precisely because we are forced to design intelligently.
Transparency will also matter. Standard reporting on power usage effectiveness, water usage effectiveness and carbon intensity should become normal practice. Clear metrics build trust. They also create performance incentives that reward operators who innovate in efficiency.
Building an intelligent AI infrastructure
The UK does not need to build the world’s largest AI infrastructure. It needs to build the most intelligent one.
That means recognising AI infrastructure as part of critical national systems. It means aligning planning reform with grid reform and skills development. It means treating digital connectivity as core infrastructure. It means embedding community benefit and local engagement into project design from the outset. And it means reforming energy markets so that clean generation translates into stable, competitive prices.
“The UK does not need to build the world’s largest AI infrastructure. It needs to build the most intelligent one.”
The alternative is not environmental protection. It is strategic drift.
If investment flows elsewhere because we fail to address volatility and integration, the UK will still consume AI services – but the economic value, jobs and technical capability will sit offshore. Climate impact does not disappear simply because infrastructure is built in another jurisdiction. It merely moves.
Britain has the research base, regulatory credibility, and institutional depth to lead in Green AI. But leadership will not emerge automatically. It requires deliberate system design.
The climate concerns being raised are legitimate. They are precisely why we must get this right.
AI can strengthen energy systems, optimise grids, improve public services, and accelerate scientific discovery. But only if the infrastructure that powers it is designed with discipline, transparency, and integration.
This is not a choice between growth and responsibility. It is a test of whether we can deliver both. UKAI believes that the UK can and should be a leader in delivering that responsible growth.