The UK Government has announced plans to cut electricity prices for AI data centres in central Scotland by £24 per MWh, claiming that the measure will be “cost-neutral” for households. It is an attractive line for ministers who want to present the UK as a global hub for artificial intelligence while reassuring the public that they will not bear the cost. But the government’s narrative collapses under basic scrutiny. Far from being neutral, the policy uses savings that rightfully belong to bill payers and redirects them to some of the world’s most energy-hungry commercial operations.
To understand what is happening, we need to start with constraint payments.A little-known mechanism that already adds billions to the cost of electricity every year. When Scotland produces more wind power than the grid can carry south into England, the system operator pays wind farms to switch off. That money doesn’t come from government budgets; it comes directly from consumers through their electricity bills.[1] In theory, if constraint payments fall, household bills should fall with them. But under this new AI support plan, any reduction in constraints will not be returned to the public. Instead, those savings will be diverted to fund the discounted electricity offered to data centre developers. This is a subsidy by another name, even if the government avoids using that word.
Government officials argue that the discount is acceptable because the presence of large data centres will “soak up” surplus renewable energy on windy days. But even if that were true, and the reality is much more complicated ,the savings that emerge from doing so still originated from public payments. They are not free money. They are prepaid by millions of households who have been absorbing constraint costs for years. Reallocating that benefit to commercial AI operators means consumers never see the price reduction they are owed.[2]
There is also a fundamental flaw in the assumption that data centres will rely primarily on surplus Scottish wind. AI data centres do not run intermittently. They cannot stop and start based on the weather. Their power needs are continuous, often 24 hours a day, seven days a week. When wind output is low, or when Scottish export capacity is not constrained, they still draw power from the grid. Yet they continue receiving the same discounted price because the policy is based on system-level avoided costs, not the real-time energy mix they consume.[3] In other words, the discount applies even when the data centre is using normal, non-surplus electricity. The public subsidy is baked in, regardless of conditions.
There is an even deeper issue hidden inside this proposal: the enormous grid reinforcement required to host data centres of this scale. A 500-600MW AI campus such as the proposed Ravenscraig development will require new transformers, new substations, reactive power equipment and major upgrades to the surrounding transmission network. These works cost hundreds of millions of pounds, and under Ofgem rules, they are not paid by the developer. They are “socialised” meaning they are smeared across the bills of every household and business in the country.[4] The public pays for the infrastructure, while the developer benefits from the connection.
As more data centres are built, the strain they place on the system increases. AI workloads do not flatten peaks; they intensify them. When demand surges, balancing the system becomes more expensive. The costs of holding backup generation, stabilising voltage and maintaining frequency security all rise in tandem with heavy, constant loads. None of this is absorbed by the data centres themselves. It is pushed into the standing charge, the transmission charge and the distribution charge.The three elements of the bill that ordinary consumers cannot avoid, cannot control and cannot reduce through personal behaviour.[5] The public ends up paying for the privilege of hosting a power-hungry industry.
Despite these facts, the government continues to insist the policy is cost-neutral. This hinges on a financial sleight of hand: treating avoided constraint payments as if they were new revenue created specifically to fund AI. But these “savings” are not new at all. They are reductions in a cost the public already faces. If constraints fall from £1.2 billion to £900 million, the £300 million saving should return to the bill payer. Under this policy, however, that saving is used to subsidise electricity for a corporate sector. It is a redistribution of public money disguised as innovation policy.
This distortion will also reshape the wider electricity market. If AI operators enjoy significantly lower electricity prices in Scotland, everyone else becomes a relatively higher payer. Including manufacturers, small businesses, public services and households. The electricity system is not being made cheaper; it is being rebalanced to favour one government-backed industry at the expense of everyone else. Industrial competitiveness suffers, while residential bills remain stubbornly high. The public is being asked to accept permanently expensive electricity so that AI developers can access cheap power that the system cannot naturally provide without subsidy.
The final question is simple: who benefits? Under the government’s proposal, AI developers enjoy reduced electricity costs, taxpayer-funded grid works, preferential pricing and political sponsorship. In return, households get nothing but the same high bills, deeper infrastructure charges and additional system stress. The claim of cost-neutrality is nothing more than political smoke. When a discount is created by diverting public savings into private hands, the public is subsidising the outcome, whether ministers admit it or not.
The truth is clear. The UK’s electricity system is already broken by years of mismanagement, extreme reliance on intermittent renewable energy and soaring network costs. Instead of fixing the system, the government now plans to use it as a piggy bank to fund the AI boom. Calling this “cost-neutral” insults the public. Every pound of saved constraint cost that goes to an AI developer is a pound that does not go back to a family struggling with energy bills. It is a transfer of wealth from households to corporations, hidden behind technical language and the promise of economic opportunity.
Ministers may insist this is a harmless step towards making Britain a global centre for artificial intelligence. But until the public is protected from rising charges, and until genuine transparency is applied to how the energy system is funded, one fact remains unavoidable: bill payers will subsidise AI, because the government has decided they must.
Footnotes
[1] UK constraint payments exceed £1.2 billion per year, paid through consumer bills via balancing and network tariffs.
[2] Ofgem’s charging framework directs all balancing and constraint costs onto end users through BSUoS, DUoS and TNUoS charges.
[3] AI data centres require continuous baseload power and cannot operate solely on intermittent surplus energy.
[4] Transmission reinforcement costs for large new demands are socialised across all users under Ofgem-approved network charging methodology.
[5] Balancing requirements — including voltage control, reserve generation and frequency stabilisation — flow directly into the standing charge, transmission charges and distribution charges.
Shane Oxer — Campaigner for fairer and affordable energy
SAY NO TO THE DESTRUCTION OF OUR LAND 
Leave a comment