Your Data Centre Has Become a Power Project - Whether You Planned It or Not

Data centre development was straightforward: secure land, obtain planning consent, appoint a connection adviser, procure a grid connection, and contract for power.

At 20MW, that model can still work. At 200MW, 500MW or beyond, it becomes dangerously incomplete.

The issue facing UK data-centre investors is not simply that grid connections take too long. It is that many projects are still being developed as real-estate schemes with a utility connection attached, when their power requirement has already made them a utility-scale energy-infrastructure project.

The question is no longer just, “Where can we connect?” It is: what is the most financeable, consentable and deliverable route to firm power at the scale and speed the campus requires?

The limits of connection-led thinking

This is not an argument for ignoring the Distribution Network Operator (DNO), National Energy System Operator (NESO), or the transmission system. They remain essential counterparts in almost every serious scheme.

But a grid application is not, by itself, a power strategy. Its an empty pipe.

The simplicity of a 132kV or lower) distribution connection can conceal a much wider set of constraints: circuit and transformer capacity, network-security requirements, upstream transmission reinforcement, fault level, substation design, protracted land rights, delivery sequencing and the queue position of other projects. There is no universal MW limit for a 132kV circuit; the answer is highly site- and network-specific. Connection guidance recognises circuit capacities both below and above 100MVA — a useful reminder that voltage level alone does not determine what a site can actually import or export.

An investor cannot underwrite a large data-centre site against an early-stage connection indication alone. The critical issue is the credible path to energisation — and the cost, time, dependencies and risks embedded in it.

The regulatory direction reinforces this point. As of last week, Ofgem is currently consulting on a data-centre commitment fee for projects above 40MW, proposed to apply from offer acceptance until readiness to connect. It is not yet a final rule, but it reflects a changing expectation: major demand projects must demonstrate commitment, financial capacity and deliverability rather than hold speculative capacity in the queue.

You need "IPP disciplines"

Once a campus requires hundreds of MW of resilient supply, the development challenge starts to resemble a traditional Independent Power Producer project.

That does not necessarily mean the data-centre investor must own and operate a power station. It does mean the project needs to be developed with IPP disciplines from the outset:

• Power-led site selection, not simply land-led site selection.

• A grid strategy that tests transmission, distribution, private-wire and behind-the-meter routes in parallel.

• A firm-power design combining the appropriate mix of grid supply, flexible generation, BESS, low-carbon generation and - over time - potentially nuclear or SMR capacity.

• Early work on fuel or energy supply, emissions and environmental permitting, planning, EPC packaging, operability and financing, all for a power station

• A commercial structure that allocates construction, availability, commodity and regulatory risks to parties capable of carrying them.

In other words, the investor needs to develop the energy system and the digital campus as one integrated infrastructure proposition.

The US has already moved

The United States offers clear examples of where this is heading:

Microsoft’s 20-year agreement with Constellation enabled the planned restart of the 835MW Crane Clean Energy Center, formerly Three Mile Island Unit 1. Microsoft did not need to become a nuclear operator; it secured long-term access to firm, carbon-free generation through an established power owner.

At the other end of the spectrum, Fermi America is pursuing an up-to-11GW Texas AI campus built around its own multi-source energy strategy, including gas, solar, wind and nuclear. Its agreement with Siemens Energy covers up to 1.1GW of gas-fired generation equipment — power-station-scale procurement in direct support of data-centre development.

These are not plug-and-play templates for the UK. Planning, networks, markets and environmental regulation differ materially. But the strategic lesson travels: leading digital-infrastructure players are no longer treating power purely as a utility service procured after the site is chosen.

They are bringing power development into the core investment case.

As explored in CM Energy Insight’s earlier article, When Big Tech Buys the Power Station: Google’s $4.75bn Intersect Deal and the Hyperscaler Energy Arms Race, the logic is increasingly “power first”: secure the credible energy solution, then build digital capacity around it.

A different investor question

For UK data-centre investors, the practical shift is simple: Do not ask only "does this site have a connection offer?". Ask whether it has a bankable route to phased, firm, affordable and increasingly low-carbon power - on the timetable required by customers and capital.

If the answer depends on years of uncertain network reinforcement, a single constrained and contested132kV route, or an untested assumption about future capacity, then the project is not merely a data-centre development risk. It is an energy-development risk.

That risk has always been navigated by the IPPs .But it must first be recognised.

CM Energy Insight helps data-centre developers and investors assess, structure and deliver utility-scale power solutions before grid uncertainty becomes a stranded-site risk.