AI Is Creating a New Energy Problem: Who Will Pay for the Power?

AI might be the biggest technology story of the decade. But behind every ChatGPT query, AI-generated image and automated business process is something much more basic: electricity.

And AI is going to need a lot of it.

As technology companies race to build bigger and more powerful AI systems, data centres are springing up to provide the computing power behind them. The problem is that these facilities aren't just large buildings full of computers. They can consume huge amounts of electricity - 24 hours a day, 365 days a year.

So, as the UK's AI ambitions grow, a fairly obvious question is starting to emerge:

Where is all the power going to come from - and who is going to pay for it?

AI is hungry for electricity

The numbers are already getting people's attention. The International Energy Agency estimates that data centres around the world consumed around 485 TWh of electricity in 2025.

By 2030, that could almost double to around 950 TWh.

And AI is a major reason why.

AI-focused data centres are growing particularly quickly because training and running advanced AI models requires huge amounts of computing power. Thousands of high-performance processors can be running continuously, generating a lot of heat and requiring equally significant cooling systems.

In other words, the AI revolution may be digital - but its energy requirements are very physical.

The UK wants to be an AI leader

The UK government clearly wants a piece of the AI boom.

There is a push to attract investment, expand computing capacity and establish AI Growth Zones around the country, which could mean jobs, investment and economic growth.

But there is a catch.

Data centres need power…lots of power.

And connecting a huge new electricity user to the grid isn't as simple as plugging in another office building.

New substations, cables and other network infrastructure may be required and in some cases, developers can face lengthy waits for a grid connection.

That's a problem when the technology industry wants to move quickly.

The grid is becoming part of the AI race

We often talk about AI as a race between technology companies, but increasingly, it's also becoming a race for electricity and grid capacity.

A data centre can potentially be constructed much faster than the electricity infrastructure needed to supply it.

The scale of the challenge is already clear: in March 2026, the government said applications for transmission-level demand connections had grown by 460% in just six months.

At the same time, the UK is trying to electrify more of the economy, increase renewable generation and move towards net zero.

That means there are more and more demands being placed on the electricity system at the same time.

And that brings us back to the big question:

Who pays?

Will some of those costs be shared across the wider electricity system?”?

This is where things get interesting for businesses outside the technology sector.

When a massive new electricity user connects to the grid, there can be significant costs associated with providing the infrastructure it needs.

There is also the wider question of what happens when electricity demand rises rapidly.

More demand doesn't automatically mean higher prices - the UK can build more generation, increase renewable capacity and improve the grid.

But if demand grows faster than supply and infrastructure, the pressure on the electricity system could increase.

And businesses will be watching closely to see how those costs are ultimately distributed.

Will large energy users pay more of the infrastructure costs they create?

Or will some of those costs be shared across the wider electricity system?

That's a debate we're likely to hear much more about.

It's not just about the electricity bill

There's another issue too: where does all this power come from?

The UK's AI ambitions are being developed alongside ambitious targets for decarbonising the electricity system.

That creates an obvious tension.

If some data centres turn to fossil-fuelled generation to meet demand while waiting for grid capacity, the environmental benefits of the UK's transition to cleaner energy could be undermined.

There are also questions around water use, cooling and the wider environmental footprint of these facilities.

The government has already been looking at the impact of data-centre growth on energy consumption, highlighting just how important this issue is becoming.

What does it mean for businesses?

Most businesses aren't going to suddenly find themselves competing directly with an AI data centre for electricity.

But the growth of energy-intensive industries is something worth watching.

For businesses, the energy market is already much more complicated than simply finding the lowest unit rate.

Wholesale prices, network charges, government policy, grid capacity and the changing generation mix can all affect the cost of electricity.

That makes energy procurement and timing increasingly important.

If your business is approaching a contract renewal, understanding what's happening in the wider market could be just as valuable as comparing today's prices.

The bigger picture

There's no question that AI could bring enormous benefits to the UK economy.

But AI doesn't run on algorithms alone.

It runs on servers, servers need cooling and servers and cooling systems need electricity.

As the AI industry grows, the UK's ability to provide enough affordable, reliable and low-carbon electricity could become one of the factors that determines how quickly that growth can happen.

The AI boom might look like a technology story.

But increasingly, it's an energy story too.

Ironically, this article about the energy-hungry world of AI was written with a little help from ChatGPT….