In November 2025, we wrote about the return of the data centre to centre stage – Back to the Future: Why Data Centres Are Front Page News Again – and how artificial intelligence (AI) had dragged the physical internet back into view. At the time, the dominant tone was ambition: capital pouring into infrastructure, sovereign capability becoming a national priority, and AI positioned as the engine of the next economic cycle.
Three months on, the mood has shifted.
Recent reporting by The Times and The Daily Telegraph has moved the debate from projection to quantification. For the first time, the scale of grid demand being requested by prospective UK data centres has been disclosed publicly, and it is forcing a more grounded conversation.
According to Ofgem, around 140 data centre projects are seeking grid connections, representing approximately 50GW of peak capacity. On a recent day, Britain’s total peak demand stood at roughly 45GW.
Even allowing for projects that may never proceed, the signal is clear: the next phase of AI infrastructure is not incremental. It is system-scale.
For context, Hinkley Point C, the UK’s first new nuclear power station in a generation, will produce 3.2GW when fully operational. The comparison is not rhetorical. It highlights the asymmetry between demand ambition and generation reality.
This is not an argument against AI. It is an acknowledgement that AI is no longer simply a software strategy – it is also a major energy strategy. And that changes the executive calculus.
From Cloud Abstraction to Physical Constraint
For the past 15 or so years, “the cloud” has enabled organisations to abstract away physical infrastructure. Compute felt elastic. Storage felt infinite. Location seemed secondary. Artificial intelligence has broken that abstraction.
Training and running large models demands high-density compute, specialist hardware, liquid cooling systems and resilient, always-on power. These are not invisible services. They are physical systems operating at extraordinary scale.
What has changed in the public discourse is not the existence of this demand, but its visibility. The numbers are now concrete. Policymakers are responding. Parliamentary committees are asking questions. UK Labour MP Toby Perkins has called for a “national conversation” about the implications of data centre growth.
For tech leaders, the important shift is that AI capacity is now inseparable from grid capacity.
The Case for Acceleration
There are compelling reasons to press ahead.
Economic Competitiveness and Sovereignty
AI capability is rapidly becoming a proxy for national competitiveness. Jurisdictions that host infrastructure attract research clusters, advanced manufacturing, fintech, life sciences and the wider innovation ecosystem that follows digital capacity.
If the UK under-builds while others accelerate, the risk is not simply slower AI adoption. It is diminished strategic influence.
For boards, proximity to infrastructure affects latency, regulatory alignment and resilience. This is not abstract geopolitics; it is operational positioning.
Productivity and Growth
AI promises measurable productivity gains across sectors: predictive maintenance in industry, automation in professional services, optimisation in logistics, acceleration in drug discovery and diagnostics.
If realised at scale, these gains could offset demographic headwinds and enhance economic output. In that context, the energy required to power AI may be viewed as long-term investment rather than short-term burden.
Catalyst for Energy Innovation
There is also a credible argument that hyperscale demand can accelerate clean generation.
Globally, major technology companies are underwriting renewable projects, exploring advanced nuclear, geothermal and long-duration storage, and investing in alternative backup systems. Large, creditworthy demand can de-risk projects that might otherwise struggle to secure financing.
In this framing, AI infrastructure becomes a forcing function for grid modernisation.
The Case for Caution
Yet the counterarguments are equally substantive.
Decarbonisation Tension
The UK operates under legally binding carbon budgets and ambitious clean power targets. If incremental data centre demand is met through extended gas generation, new peaker power plants or fossil-fuel-intensive backup systems, the tension becomes explicit. AI growth could unintentionally slow wider decarbonisation progress.
For enterprise leaders operating under Scope 3 scrutiny, infrastructure carbon intensity is not someone else’s problem. It flows through supply chains and into reporting obligations.
Cost Allocation and Public Perception
Recent coverage has also raised a politically sensitive issue: who pays for the grid upgrades?
If network reinforcement costs are socialised across billpayers, households effectively subsidise hyperscale expansion. In parts of the United States, wholesale electricity prices near major data centre hubs have risen sharply, fueling public backlash.
Even if the UK adopts a model where developers fund their own connections or pay prioritisation fees, perception will matter. In a period of sustained cost-of-living pressure, optics are powerful.
Grid Crowning-Out
There is also the risk of displacement.
If large data centre projects dominate connection queues, renewable generation, industrial electrification and EV infrastructure may face delays. Paradoxically, infrastructure intended to power the digital transition could slow the broader energy transition.
Asset Specificity and Concentration Risk
Many AI-optimised data centres are highly specialised. If tenant demand shifts or a hyperscaler alters its strategy, these facilities may not be easily repurposed. Concentration risk becomes material. For boards, this translates into vendor dependency exposure and location risk.
The Capital Markets Dimension
A further development underscores how entrenched this expansion has become.
Recent reporting in the Financial Times highlights that developers are now seeking credit ratings for data centre projects while facilities are still under construction. Agencies including Fitch Ratings, Moody’s and S&P Global Ratings have expanded coverage of digital infrastructure, enabling tens of billions of dollars in investment-grade debt to flow into AI-driven build-outs.
Most of these projects are underpinned by long-term leases with hyperscale tenants whose creditworthiness effectively anchors the rating.
This is no longer speculative infrastructure. It is institutional capital underwriting long-term expansion.
That matters because once projects are financed and syndicated into insurance portfolios, pension funds and structured credit markets, expectations harden. Returns must be delivered. Delays, grid constraints or sustained energy price volatility do not simply inconvenience developers; they reverberate through lenders, balance sheets and refinancing structures.
AI infrastructure is becoming embedded not only in national energy systems, but in global credit markets.
Electrons now influence bond spreads.
The Executive Reality
For technology leaders, the question is not whether AI adoption will continue. It will.
The more relevant question is whether organisations are modelling the energy implications with the same rigour applied to cybersecurity, regulatory compliance or supply chain resilience.
That requires reframing assumptions:
- Location strategy must account for grid capacity and connection timelines.
- Procurement strategy may increasingly involve power purchase agreements and renewable alignment.
- Financial planning must consider exposure to wholesale price volatility.
- ESG reporting must incorporate infrastructure carbon intensity.
- Vendor selection should evaluate energy efficiency, PUE performance and long-term sustainability strategy.
Energy is no longer a background utility in digital transformation. It is a strategic variable.
The Bigger Question
Are we sleepwalking into a grid crisis?
The phrase is deliberately provocative. The numbers disclosed this month do not guarantee a crisis. Many proposed projects will not proceed. Hardware efficiency improvements may moderate load growth. Policy reform may accelerate grid expansion and clean generation.
But what has changed is visibility.
In the early 2000s, the risk was overcapacity driven by speculative optimism. Today, the risk is that demand, capital-backed, credit-rated and strategically prioritised, outpaces infrastructure readiness.
The first data centre boom asked whether the internet was economically viable. The second asks whether artificial intelligence is energetically sustainable.
Capital is available. Demand is accelerating. The technology is advancing at extraordinary speed.
But grids expand slowly. Generation projects require time. Public tolerance for unintended cost transfer is finite.
For tech leaders, the implication is clear:
AI strategy and energy strategy are now inseparable.
The organisations that recognise this early, and integrate energy realism into digital ambition will not only mitigate risk, they will help shape a more resilient foundation for the next phase of the digital economy.
The question is not whether AI will transform business.
It is whether we will build the physical and financial infrastructure to support it responsibly or discover too late that ambition has outrun capacity.
One thing that hasn’t changed is the value of partnering with a strategic data centre provider who understands both the technical and operational realities of AI infrastructure including the energy, capacity, and sustainability dimensions that now define modern digital strategy.
We design infrastructure that is secure, scalable, and energy-conscious, helping your business operate reliably while navigating the evolving energy landscape.
In a world where digital ambition meets physical limits, having the right partner is what ensures opportunity doesn’t outpace capability.
Chat with us today about our London and Switzerland data centres.
Photo by Zac Wolff on Unsplash
And don’t forget to SIGN UP to our NEWSLETTER for up to date industry news and insight delivered straight to your mail box.
Comments are closed.