18th August 2026
There is a new electricity consumer on the block and it is potentially enormous. Bill takes a closer look at this big change that is already underway
This new phenomenon does not have a television, an electric shower or a heat pump. It doesn't worry about turning the lights off or whether the electricity meter is running too quickly. It is the data centre.
Behind the explosion in artificial intelligence, cloud computing, streaming, digital services and online storage lies an increasingly hungry electricity industry. Giant buildings packed with computer servers can consume as much electricity as a small town, and the biggest proposed developments can require hundreds of megawatts of power.
That raises an awkward question for ordinary electricity consumers.
If Britain needs to build more electricity generation and massively expand its electricity grid to accommodate the AI revolution, who is going to pay for it? Ifthere is a possibility that yu an I are going to foot partof the bill action need to be taken now.
And yesterday's announcement from the Scottish Government shows that this is no longer simply a theoretical questionand Scotland is already preparing for the data-centre boom.
On 17 August, the Scottish Government announced a new planning direction covering data-centre developments with a power capacity of more than 50 megawatts.
Planning authorities will now have to notify Scottish Ministers of new applications within seven days of validation. The Scottish Government says the purpose is to monitor the pipeline of large data-centre proposals and consider their implications for wider infrastructure.
The announcement also revealed that a planning appeal for a proposed green data centre in Edinburgh has been recalled for a ministerial decision, while Environmental Impact Assessments have been directed for proposed developments in Fife and Edinburgh.
Scottish Government data centre planning direction is significant.
The Scottish Government clearly sees opportunities in the sector. Data centres can bring investment, construction work, highly skilled employment and potentially major economic activity.
But it also recognises that developments of this scale have consequences beyond the boundary of the site on which the computers sit.
One of those consequences is electricity as AI needs an extraordinary amount of power. Artificial intelligence is not weightless and every time we ask an AI system to produce an answer, create an image, analyse information or perform a complicated calculation, computers somewhere have to do the work.
Those computers require electricity and unlike a factory that might operate one or two shifts a day, a major data centre can operate around the clock.
The scale of Britain's potential demand is already attracting the attention of Ofgem. The regulator has described a rapid growth in demand-connection applications associated with data centres and is now consulting on reforms specifically designed to manage that growth.
Ofgem has even proposed a Data Centre Commitment Fee. Under the proposal, large data-centre developments accepting a grid connection offer could have to pay between £237,500 and £712,500 per megawatt. The money would be refunded if the project reaches energisation, but could be forfeited if the developer leaves the connection queue early.
Why introduce such a system? Because there is a problem.
Britain has limited grid capacity in the places where it is most needed, and speculative projects can occupy connection capacity for years.
Ofgem says it is reforming demand connections to deal with the fast growth of data centres and make better use of the electricity network.
But what happens to our electricity bills? This is where the debate hits all of us. A data centre will obviously pay for the electricity it consumes.
But that is only part of the cost as suppose a company wants to build a huge data centre in an area where the existing electricity network cannot supply the required power.
New substations may be needed and new high-voltage cables may be required. Transmission capacity may have to be increased and almost certainly additional generation could be needed.
The electricity system may also require more flexibility and backup capacity to cope with changing demand.
All of that costs money.
And Britain already knows that expanding the electricity network can affect consumer bills.
Ofgem's approved electricity and gas network investment programme is expected to increase network charges, with the electricity element of the previous investment package estimated at around £74 per household by 2031. At the same time, Ofgem argues that investment in the electricity grid can ultimately reduce costs by allowing cheaper electricity to reach consumers and reducing reliance on expensive imported energy.
That illustrates the complication that grid investment isn't necessarily bad for consumers. In fact, failing to invest could be considerably worse.
But when new infrastructure is required partly because of enormous new industrial electricity demand, we should be asking whether the companies creating that demand are paying an appropriate share of the cost.
The danger of making households subsidise AI and this is where policymakers need to be careful.
Imagine a data centre requiring hundreds of megawatts of electricity. The developer pays for its building and equipment. It pays for the electricity it consumes. But suppose additional network investment is spread across electricity consumers generally.
Then someone living in Caithness, Glasgow or Dumfries could end up contributing towards the cost of infrastructure that is helping a multinational technology company operate a huge computing facility.
That would be difficult to justify if the household received no corresponding benefit.
The same principle applies to electricity generation. If Britain has to construct additional generation because demand from data centres is growing rapidly, the cost of that investment has to be recovered somehow.
The question shouldn't simply be: "Can Britain afford to have data centres?" It should be - "Can Britain afford to have data centres without transferring their infrastructure costs onto everyone else?"
Scotland has an unusual opportunity and there is, however, another side to this argument.
Scotland has enormous renewable electricity resources as we have wind farms producing electricity, particularly in areas such as the Highlands and northern Scotland, where the electricity network has sometimes struggled to transport all the power being generated.
That creates an interesting possibility in that rather than seeing data centres simply as another burden on the electricity system, Scotland could potentially use them as a major source of new electricity demand located close to renewable generation.
If a data centre can be located in the right place, with adequate grid capacity, it could consume electricity that might otherwise have been constrained or curtailed. That could be particularly interesting for the Highlands.
But location matters. Putting a giant data centre into an already constrained part of the network and then spending billions strengthening the grid to supply it is very different from locating major electricity demand alongside plentiful generation and available network capacity.
This is why the Scottish Government's new planning direction is potentially important. The question isn't simply whether Scotland wants data centres it is also where they should go, how large they should be and what infrastructure they will require.
What about us in Caithness? This is where the debate becomes particularly relevant locally. Caithness has something many parts of Britain would like more of — substantial renewable energy resources and major existing and planned electricity infrastructure.
The Pentland Firth and surrounding area have enormous renewable-energy potential. Dounreay also provides a unique combination of an existing nuclear-energy heritage, engineering skills, electricity infrastructure and a highly skilled workforce.
Could that eventually make northern Scotland attractive for energy-intensive industries, including data centres? The jury is out the posibilities.
But there is an important warning that simply attracting a data centre is not an economic development strategy.
The community needs to ask what it gets in return.
How many permanent jobs would be created and how much business would go to local companies?
Would there be investment in local infrastructure and would the development help utilise electricity that would otherwise be constrained?
Would the company contribute significantly towards the infrastructure required to connect it?
And most importantly would the economic benefit remain in the Highlands, or would Scotland simply provide the land and electricity while the profits flow elsewhere?
These are questions worth asking before planning permission is granted, rather than afterwards.
Britain's electricity demand is changing and there is another reason why this debate matters.
Britain is already attempting to electrify much more of its economy. We are seeing electric vehicles increasing electricity demand. Aoher area coing on rapidly are heat pumps which will increase electricity demand.
Industry is expected to electrify and hydrogen production could require enormous quantities of electricity.
And now AI and data centres are adding another potentially massive source of demand. The electricity system therefore faces a fundamental challenge.
Britain needs much more electricity — and it needs it to be affordable.
That means building generation, strengthening the grid and making better use of the electricity we already produce.
But there is a danger that policymakers become so enthusiastic about attracting new investment that they overlook the cost of providing the electricity infrastructure necessary to support it.
The American experience provides a warning to us herein the UK.
Parts of the United States have experienced an extraordinary data-centre boom, with electricity demand rising sharply in areas where AI and cloud-computing facilities are being built.
That has produced an increasingly familiar argument for us to examine. Should ordinary households pay for infrastructure required largely because enormously wealthy technology companies are demanding vast quantities of electricity?
Britain should learn from that debate rather than waiting until the problem becomes politically difficult.
Data centres could actually help consumerand it would be wrong to conclude that data centres are necessarily bad news for electricity customers.
They could bring substantial benefits.
A well-located data centre could create investment and skilled employment, provide a major new customer for renewable electricity, help use electricity that might otherwise be curtailed, encourage investment in new generation, contribute towards grid infrastructure, generate business rates and other economic benefits and establish new technology industries outside Britain's traditional economic centres.
There could even be opportunities to use waste heat from some facilities, although that depends heavily on location and the availability of suitable nearby heat demand.
So the issue isn't data centres versus ordinary consumers as it is more about getting the economics right.
The principle should be simple and we should say, " If a new industry requires enormous amounts of electricity infrastructure, it should make a fair contribution towards the cost of providing it."
That doesn't necessarily mean making data centres pay every penny of every network investment.
The grid is a national asset, and investment can benefit many other users as well.
But neither should households automatically be expected to carry costs that would not otherwise have existed.
Ofgem's proposed commitment fee is one indication that regulators are beginning to recognise the problem.
The Scottish Government's new planning direction is another recognising Scotland needs to decide what it wants from the AI revolution as there is a bigger question here than electricity bills.
Scotland has spent years trying to attract investment and create new industries. Data centres could become an important part of that future.
But Scotland has also spent years trying to increase renewable electricity generation and decarbonise its economy.
The two objectives could potentially fit together very well.
A data centre located close to abundant renewable generation could be considerably more attractive than one requiring huge amounts of new transmission infrastructure.
That is why the Highlands should not simply ask:
"Do we want a data centre?"
We should be asking, "What sort of data centre would actually benefit the Highlands?"
And that means looking at electricity, grid capacity, employment, rates, infrastructure, local supply chains and long-term economic benefits together.
The Scottish Government's announcement this week suggests that the issue is finally being treated as something that requires strategic planning.
That is probably a good thing.
Because the AI revolution is coming whether we like it or not and we should make sure it works for everyone. The important question for electricity consumers is whether we end up paying the electricity bill for it.