Powering AI without breaking the grid.

The grid’s largest new customer.

For most of the past century, electricity demand in advanced economies grew slowly and predictably, and grid planners could see it coming. AI has broken that pattern. In a growing number of systems, Data Centres are now the largest single source of new load, arriving in hundred-megawatt blocks with two-year timelines against grids planned over ten. The response has been predictable too. Dublin restricted new connections. Amsterdam paused them. Frankfurt and Northern Virginia ration them. Communities that once welcomed a Data Centre as clean, quiet employment now ask what it will do to their bills.

They are right to ask. A gigawatt of new load, unplanned, is paid for by everyone else on the network, through reinforcement costs and tariffs. The industry has treated the grid as a utility to be consumed. It is about to discover that the grid is a partner to be earned.

Load without generation is a liability.

A developer who arrives with only demand is asking a community to fund its ambition. That is the honest description, and it is why the yes is getting harder to obtain. The alternative is to arrive as a net contributor: with new generation, with the ability to flex, and with heat that goes somewhere useful. This is not corporate social responsibility. It is the price of a connection.

“If you turn up with a gigawatt of demand and nothing else, you are not a customer. You are a problem the grid operator now has to solve.”Jamie MacDonald-Murray, Chairman & CEO, Eppur

Three things to bring.

Generation first. Every Eppur campus is renewable by default, backed by power purchase agreements with grids and generators we know first-hand, and wherever we can we bring new clean capacity to the network alongside our load. Additionality answers the question the industry has dodged for years: whose clean power is it, anyway?

Flexibility second. A Data Centre is a controlled energy system: power in, heat out, intelligence in between. Designed properly, it can be one of the most flexible large loads on the grid. Batteries on site, training workloads that can shift by the hour, cooling that can ride through a peak. A campus that can drop a hundred megawatts for the operator on a cold evening is worth more to the system than one that cannot, and increasingly it will be paid for the difference.

Heat third. An engine running at this intensity produces heat by doing its job. We design that heat to be reused, into district networks and industrial offtake wherever a partner exists, rather than vented into the sky. Our design standards are public: a power usage effectiveness of 1.20 or better on every site, water usage disclosed per facility per year, emissions reported by scope. Receipts, not adjectives.

“We connected gigawatts of wind and solar by being the easiest customer the grid operator had that year. Bring generation, give them flexibility when they ask for it, and the connection date moves in your favour. It is not complicated. It is just work.”Peter King, Co-founder, Eppur

Why regulators say yes.

Consent follows credibility. The operators and planning authorities we work with have seen every version of the brochure. What moves them is a developer who publishes its numbers, brings generation to the table, and has done this before. Our founders spent two decades connecting solar and wind to grids across the United Kingdom, Europe and Brazil. Those relationships are the reason our sites secure power that others cannot, and the reason we build the way we do.

“We spent twenty years putting clean power onto grids. We are not going to spend the next twenty taking it off.”Jamie MacDonald-Murray, Chairman & CEO, Eppur

The AI era needs an extraordinary amount of electricity. It does not need to break anything to get it. It needs developers who understand that the grid is the product, and the building is what sits on top.