To Cut Electricity Bills, Britain Needs to Use More, Not Less

What's Next

Executive Summary

Britain's high energy costs relative to peers are driven by expensive electricity. Gas prices are high across Europe, but they are not what sets Britain apart. British domestic electricity prices are the fourth highest in the EU14+GB group and non-domestic the highest, while domestic gas prices are the lowest. Great Britain has the highest electricity-to-gas price ratio in the group, at 4.2, around 60% above the peer median. Our model shows this ratio rising by around a fifth by 2035.

Britain’s cost of electricity supply and networks is level with most peers. Bills are not higher because it costs more to produce electricity here, but because of taxes and levies, as well as the pass-through of gas prices.

The sustained rise in the cost of building and running the system comes from a combination of falling demand and increasing fixed costs, driven by generation and grid build-out. An 11% fall in demand since 2015 explains 39% of the rise in electricity system cost. Gas still matters, especially during an energy price shock, but it is not the sole cause of the 2015–24 rise in consumer electricity bills. Wholesale prices explain 38% of that rise; but network and policy costs explain 45%.

These are related; GB is in a negative demand/price spiral on electricity and energy more broadly. This is only likely to get worse. There are no easy choices to solve this, all involve difficult trade-offs.

To Cut Electricity Bills, Britain Needs to Use More, Not Less | Exhibit 1

If no action is taken, consumer electricity bills will rise in nominal terms by £115 (13%) to £996 in 2035, in our central scenario (Exhibit 1). Fixed network costs drive 86% of this. This assumes no further shocks and depends on demand growing in line with generation. In real terms, bills fall slightly (by around 5%) over the next decade in our central scenario.

Continued decline in demand would drive bills up by a further £149 in 2035, a 15% rise compared to our central scenario. GB’s grid electricity demand has consistently grown more slowly than forecasts and generation build.

Our model shows that bills are becoming increasingly fixed at the bottom end. If the wholesale gas price fell by a third it would only cut bills by 3% in 2030. Yet bills remain exposed to upside price shocks, though less so over time.

No single lever, on its own, can significantly reduce bills. We tested eight levers across four groups: how costs are allocated, how the system is run, how the system is built and maintained, and how it is shaped for the long term. Only the first and the last have a material impact in the short-to-medium term. We propose a package of five levers (Exhibit 1).

These two levers must work in tandem to deliver meaningful bill reductions. Costs could be temporarily reallocated away from bills to support measures that unlock demand over the medium to long term. Together, these can reduce bills by £174 a year in 2035 relative to our central case. The aim is to kickstart a positive spiral in which costs are recovered across a larger usage base when the system’s marginal cost is lower.

We set out four options to temporarily remove costs from bills in the short term in a fiscally sustainable way. They would work best as a package:

moving social costs, such as the Warm Home Discount, from domestic bills to general taxation (£15 a year), and doing the same for Renewables Obligation (RO) and Feed-in Tariffs (FiT) on non-domestic bills;
slowing network depreciation to match real asset life (moving from 45 to 55 years), worth £10 a year;
utilising the National Wealth Fund or a similar mechanism to defer capital-related network charges from bills, an upper bound impact of £153; and
financing new infrastructure more cheaply from the outset, worth up to £33 a year.

Industrial demand, in the form of data centres, is the most likely and most economically additive driver of demand. This is for four reasons: 1) Consumer demand is constrained. 2) Current forecasts already bake in ambitious adoption of EVs and heat pumps. 3) Domestic demand is mainly demand switching from other fuels, so not net economically additive. 4) Other industrial sources of demand have been declining for decades, while data centres are growing strongly.

Delivering 14GW of the current data centre pipeline would cut every household's bill by £35 by 2035, if done in a grid-positive way and without requiring significant new generation. Britain could let projects connect faster where they locate intelligently, operate flexibly and pay the costs they cause. Four reforms can do this: non-firm connections in return for limited curtailment; a national market for powered land; developer self-build; and behind-the-meter power as a bridge to full grid capacity. Operators should also commit to specific local improvements to secure the support such a rollout needs.

Britain is at a crossroads when it comes to energy costs. Without action, bills will keep rising. That is largely a conscious choice: fixed costs have been loaded onto billpayers when they can least afford them. Deferring them is progressive and logical given the lower marginal cost system Britain is building towards. Counterintuitively, the requirement is that Britain use more electricity, not less. Britain is already investing heavily in its grids, but the scale and timing of electrification are uncertain. Large new loads from data centres can help cover that investment. In doing so, it can also reduce costs for consumers and businesses, support wider economic growth, and enable the UK’s AI ambitions. A win, win, win.

Context: Electricity Costs Drive Britain’s High Energy Bills

Britain’s high energy costs are a structural problem for the country. They raise the cost of living, the cost of doing business and harm economic growth. At the core of this is high electricity prices for domestic and non-domestic users.

To Cut Electricity Bills, Britain Needs to Use More, Not Less | Exhibit 2

Energy costs cover electricity and gas, but there are two reasons why we focus on electricity prices:

British electricity is far more expensive than peers’; gas is not (Exhibit 2) – British domestic electricity prices were the fourth highest of the EU14+GB, and non-domestic prices were the highest of the group. Gas tells the opposite story: domestic gas prices were the lowest in the group and non-domestic prices sat just below the median.¹
Britain has the highest electricity-to-gas price ratio in the group. On domestic prices in the second half of 2025, Britain’s ratio was 4.2x, against a peer median of 2.7x – around 60% higher.²

This is not to say gas is irrelevant, far from it. Britain, like much of Europe, remains exposed to wholesale gas price shocks. Following the 2022 shock, prices remain 35% higher than before. This feeds through to electricity prices whenever gas is the marginal plant.

Right now, Britain is not an outlier because it costs more to generate electricity, but largely for two other reasons:

Taxes and levies on bills – in 2024 countries such as the Netherlands and Austria had similar underlying energy network and supply costs (24p/kWh) but net negative levies and taxes on bills.
Gas price pass-through – countries such as France and Spain have similar taxes and levies, but gas sets their electricity prices far less often, partly because of France’s significant nuclear generation and Spain’s large renewables generation.

This is why further North Sea exploration is unlikely to solve Britain’s high bills. British gas prices are set by the wider, interconnected European gas market, not by how much gas Britain itself produces. Domestic production is already less than half of British demand and is declining. Further exploration and utilisation could well be worthwhile for other economic or policy reasons. For example, North Sea reserves are more secure than imported supply, carry lower upstream emissions than imported LNG, and could provide a useful fiscal windfall for the government.

There is a related but separate consideration to be made about how Britain buys gas. The market was built for an era of abundant piped supply and a more benign geopolitical environment. Greater storage and more long-term contracting could damp the volatility that flows through into electricity prices. We do not consider this in detail here, but it could be a helpful additional step to mitigate exposure to gas price shocks, and the related fiscal implications, as Britain transitions towards a system that is better insulated.

Two further distinctions matter. "Bills" are what a typical household or business is charged – most relevant for affordability. "System cost" is the underlying economic cost of building and running Britain's power system. Today, the bill sits well above the system cost, and that gap has more than tripled since 2015 (Exhibit 3). The bill includes several elements that system cost strips out: electricity bought at market price rather than economic cost, meaning it captured the full force of the 2021-22 gas shock;³ policy levies and VAT;⁴ and network and retail costs measured at the domestic tariff rate, which is close to twice the system-wide average per kWh because households connect at low voltage and spread fixed costs over small volumes.⁵

To Cut Electricity Bills, Britain Needs to Use More, Not Less | Exhibit 3

So, if not solely gas, what has driven British system costs and bills higher than peers? An 11% fall in total GB demand since 2015 explains 39% of the rise in electricity system cost – more than any other factor. There is no single cause. Economic growth and electrification (early EV and heat pump adoption) both partially offset the fall, while building efficiency, structural economic change, and industrial decline were the main drags. By contrast, for bills, wholesale prices (38%) and network and policy costs (45%) have been the key drivers of the increase.⁶

Vision of the Future: Higher Fixed Costs Driving Rising Bills

What has driven bills historically won’t necessarily drive them in the future.

To understand what will, we built a new forecasting model of consumer bills across three scenarios:

Central baseline: reflects current policy and market dynamics, utilising the Department for Energy Security and Net Zero’s central demand forecast.
Low-demand case: the same assumptions but with forecast demand growth replaced by the gentle decline of the past decade.
Upside case: adjusts the central baseline for the proposed levers to reduce bills, which we set out below.

All three assume wholesale gas and carbon prices are in line with current forecasts and future prices. That assumption is uncertain and probably unlikely to hold, but it remains important. Even on relatively benign wholesale prices, relief will not arrive automatically.

In the future, falling gas prices alone will not meaningfully bring bills down. A one-third fall in the wholesale gas price would cut bills by just 3% in 2030, despite wholesale electricity prices falling 17%.⁷ Three mechanisms underpin this: carbon costs make up a growing share of the marginal cost of generation; wholesale will be only about a quarter of a household bill; and contracts for difference claw back roughly a quarter of any fall. The top-up they pay generators is the gap between a fixed strike price and the wholesale price, therefore it rises as wholesale falls. The overall effect is also not symmetric. As gas falls, wholesale shrinks as a share of the bill and further falls matter less. But as gas rises, it grows, and the bill follows more closely.

To Cut Electricity Bills, Britain Needs to Use More, Not Less | Exhibit 4

As Exhibit 4 shows, in our central baseline scenario bills are set to keep rising, but for a narrower set of reasons than in the past. Our model also suggests the ratio of electricity to gas unit prices will rise by around one-fifth by 2035. Network build-out is by far the largest driver, accounting for roughly 86% of the forecast increase.⁸ Notably, policy costs and VAT are forecast to be broadly flat or slightly declining over this period, in contrast to their significant contribution to the historical rise. The affordability challenge ahead is increasingly a network- and market-design story, not a policy-cost story.⁹ This is a nominal, not a real, rise. In real terms, bills fall slightly (by around 5%) over the next decade in our central scenario.¹⁰

To Cut Electricity Bills, Britain Needs to Use More, Not Less | Exhibit 5

This baseline itself requires demand growth. Under the low-demand case (simply continuing the demand trajectory of the past decade) bills would be around £149 higher per year by 2035 than in the baseline scenario. Meanwhile our upside case, combining further demand growth with policy reform and reallocation of network costs, could instead save around £174 (Exhibit 5).¹¹

Recent high prices have also spurred adoption of solar and home batteries. This type of on-site electricity consumption never crosses a meter, so carries none of the network, policy or capacity costs that it would if taken from the grid. This is a rational response and, for those who can afford the upfront cost, a direct route to a lower bill. But the fixed costs in the system do not disappear – they move onto everyone else, and disproportionately onto renters and households without the capital to opt out. Our model puts this at around £12 to £25 a year on a typical bill by 2035, depending on how many of those households include a battery.¹²

The negative price/demand spiral is therefore likely to worsen: bills suppress demand, which in turn raises unit costs given the high fixed-cost base. Breaking out of this spiral, and ideally reversing it, is essential.

While our model focuses on consumer bills, industrial users face a similar story. Delivered electricity prices for the average non-domestic consumer rose by 131% between 2015 and early 2026,¹³ driven principally by wholesale costs.¹⁴ Industrial demand has also fallen sharply. In 2025, it fell to its lowest level since 1983.¹⁵

Levers to Improve Affordability

We examined eight levers with the potential to improve affordability, across four groups, set out in Exhibit 6.

To Cut Electricity Bills, Britain Needs to Use More, Not Less | Exhibit 6

Exhibit 7 sets out the estimated 2035 bill and system-cost impact of a select set of levers across our groups.¹⁶ No single lever can significantly improve affordability. In many cases, the levers are interrelated and should therefore be seen as groups of interventions.

To Cut Electricity Bills, Britain Needs to Use More, Not Less | Exhibit 7

Shifting how the system is built, maintained and run has limited potential to reduce prices over our time horizon. On their own, some of these levers have limited impact:

Shaving peak demand through time-of-use tariffs worth -£6/year, -0.25 p/kWh by 2035. This figure captures only the operational wholesale effect. Flatter peaks displace the most expensive gas plant, reducing fuel and carbon costs, though around a third of that gain is absorbed by higher contract-for-difference top-ups. It deliberately excludes any reduction in generation and network build.¹⁷
Reducing the cost of capital behind CfD strike prices, worth -£3/year, -0.13 p/kWh.

One lever we have not modelled is how generation is remunerated. For example, contracting firm gas capacity directly could weaken the link to wholesale gas prices. This would trade more fixed cost for less exposure to volatility, with any revenue removed from the market needing to be replaced through contracts or capacity payments.

Shaping the system is the only way to affect system costs at scale, but it is a long-term lever. Britain has already committed to a low-carbon generation mix, and any change of direction would be a huge undertaking that would not bring bills down within five years. We are not judging whether such a shift is worthwhile. That is a separate discussion. The point is simply that it does not solve Britain's current problem of embedded high electricity costs over the next five to 10 years. Because so much of the system is financed through fixed upfront costs, there is little scope to change it once the decision to build has been made. Over the longer term, the aim must be to build neither too much nor too little. This means weighing capex – such as further grid build-out – against opex that delivers flexibility – such as new forms of storage, dynamic line rating and grid-enhancing technologies – while allowing the existing grid to serve the same need at lower cost.

For all these reasons, over the next five to 10 years, demand is the main system-shaping lever that can meaningfully bring bills down. It is also a defensive step: bills will rise further if demand continues to underperform.

The other meaningful lever, and the only one with a rapid impact, is how costs are allocated. The range is wide, depending on which costs are shifted and whether the change is permanent or temporary. Moving key policy levies onto general taxation could bring domestic bills down by £69 a year. This does not change underlying system costs, but it could be crucial to unlocking demand.

Britain therefore needs cost-allocation levers in the short term and demand-shaping levers over the medium to long term, aimed at reversing its negative price spiral. Moving costs off bills gives near-term relief, but that is not the primary aim. It is there to support demand growth, spreading fixed costs over a larger base, and then be recouped at a time when it can be better afforded.

Policies to Improve Near-Term Affordability

At its core, this is a timing problem. The grid and generation costs of transitioning to a variable, renewables-driven system are landing precisely when they are least affordable. During this transition, the potential continued pass-through from higher gas prices, combined with high fixed costs, leaves Britain’s system inflexible and exposed. This exposure will ease over time, but additional near-term bill relief can also help provide a further buffer against shocks.

The key challenge to improving short-term affordability is moving some costs off bills in a way that is fiscally sustainable, given limited room under the fiscal rules. We take these rules as a given here; however, this does result in some complex financing mechanics. It would be possible to avoid these through changes to the rules, for example by allowing a broader range of assets (such as energy infrastructure) to net off debt liabilities.

We see four possible approaches to achieve this, covering both domestic and non-domestic users. These can operate individually but are most effective when implemented together.

First, some policy costs could move onto general taxation. Social costs such as the Warm Home Discount would be more progressive if funded through tax rather than bills. This would cut bills by £13 in 2027, rising to around £17 by 2040.¹⁸ For non-domestic users, costs relating to existing assets – the RO and FiT – could be temporarily removed for all businesses, in line with the recent removal of some household costs. That would equate to roughly £800 a year for a small business using 20,000 kWh.¹⁹ Together, these measures could reach around £5.5 billion a year in fiscal cost.²⁰

Second, consumers today are potentially funding the cost of transmission assets faster than the economic life of those assets requires. Ofgem currently depreciates new additions over roughly 45 years against technical physical lives which could be closer to 55 years (based on Ofgem’s 2024 analysis).²¹ Reviewing and resetting capitalisation rates to close this gap could reduce bills by around £10 annually, with savings rising as the new asset base accumulates. This is a conservative estimate. Network assets can last even longer. Extending the depreciation period further would generate greater savings. The main obstacle is that slower depreciation weakens the cash-flow ratios on which network credit ratings depend. A review and consultation between the government, the network operators and their shareholders (predominantly pension funds) could help find a way through this. One option would be to use bridging finance or government underwriting to help maintain cash flow and credit ratings.

Third, remaining network costs could come off bills through revenue reprofiling, applied to a specific slice – such as capital-related revenue, which accounts for just over half of the increase in network costs.²² Ofgem would reduce network charges during the current high-price period, and a public financing facility delivered by the National Wealth Fund would lend to NESO or directly to the networks the cash they would otherwise have received from consumers. That loan would be a financial asset for the government, netting off debt on a Public Sector Net Financial Liabilities basis if provided at fair value. The deferred charges would then be recovered gradually from future bills later in the asset’s life and used to repay the loan. This could also be subject to affordability tests and annual caps. If applied only to domestic bills this amounts to around £5 billion per year, rising to nearly £10 billion per year if non-domestic included too. This needs to be carefully balanced. This should be done in conjunction with lenders and investors, to ensure it can be designed in a way that does not adversely impact the cost of capital for the crucial network build out more broadly.

Fourth, new infrastructure can be financed more efficiently from the start. Where public finance lowers the risk-adjusted cost of capital, National Wealth Fund funding would be available before CfD auctions and network competitions. Developers would then compete on the basis of this cheaper finance, so the benefit is locked into lower strike prices or regulated revenues, rather than retained by owners. On a like-for-like basis (comparing Ofgem's own risk-free rate with its allowed cost of debt at RIIO-3) the UK government borrows up to 150bps below an electricity transmission operator's allowed cost of debt.²³ This could generate savings of up to £33 per year, though because lending doesn’t typically occur at pure gilt rates,²⁴ the effect could be smaller.

If this approach is taken, it will be crucial to ensure that the bill reductions gained are not gradually eaten away by rising costs in other parts of the bill or system, as has often been the case historically. Publishing transparent analysis of the ongoing impact of these choices on bills could be one way to track and avoid erosion over time.

This is not an attempt to make costs disappear. It is an attempt to fund them at the right time and in the right way. However, on a pure net present value basis, this could end up costing more overall, not least because it would require additional borrowing at a time when borrowing is expensive. Even so, it could be worthwhile if three conditions hold:

First, more demand comes into the system, spreading these fixed costs across a broader base in future.
Second, now is a particularly bad time for the costs to come due. If the government expects today's investment to produce a larger and more affordable electricity system in future, it makes sense to finance a bounded share of long-lived investment over the period in which those assets will be used.
Third, a growing number of households are in energy debt and borrowing at high interest rates to manage it. Around 1.13 million consumers are in arrears on electricity bills, almost double the level of a decade ago.²⁵ For this group, the net present value of the approach is likely to be strongly positive.

This points to a wider conclusion: the approach is progressive. Lower-income households would benefit disproportionately from a cut to bills because they spend far more of their budget on energy, particularly relative to tax. Electricity has risen to around 7% of income for the poorest fifth, more than double the 3% population-wide average.²⁶

It is often argued that asking future generations to pay for today’s transition is unfair. We would argue the reverse. Costs should fall on those who will benefit most from the assets, when users are less price-sensitive and there is more demand over which to spread them.²⁷

The Key System-Level Lever: Increasing Demand

Driving up demand growth can trigger a positive spiral: more demand lowers unit costs, which supports further consumption, which reduces costs again. Both domestic and industrial users show a negative price elasticity of demand, at around –0.61 for British households²⁸ and –0.82 for industry.²⁹ Demand growth is also what makes shifting costs off bills temporary rather than permanent, by spreading them across a larger base in future. But how likely is that demand to materialise, given it has declined consistently for a decade, and where will it come from?

We believe data centres offer the most likely and most economically beneficial route to demand growth for four reasons.

To Cut Electricity Bills, Britain Needs to Use More, Not Less | Exhibit 8

First, consumer demand is more constrained than industrial demand. There is only so much additional electricity households can use.

Second, today’s baseline scenarios already include optimistic assumptions about EVs and heat pumps, the most likely sources of additional demand:

EV demand rises by 27 TWh. This equates to the number of EVs on the road in the UK rising from 3.25 million today to around 17.4 million in 2035.³⁰ Given that recent EV uptake supports this trend and that this year, the average new electric car was cheaper than the average new petrol car for the first time, we believe there is a high likelihood this is achieved. This also matters for the system, not just for demand. EVs on smart tariffs add substantial consumption without increasing peak demand, so they can add both load and flexibility.
Heat pumps are less certain: the 18 TWh increase implies a stock of around 5.5 million by 2035, up from 250,000 today, and installations remain well below the necessary trajectory. We see only a moderate likelihood of this being achieved, and it could require specific government intervention and spending.³¹
Hydrogen electrolysis demand rises by 12 TWh from near zero today.³² This looks to be very unlikely given there has been limited uptake in recent years.
Finally, the central scenario assumes base demand grows by 17 TWh over the next decade, despite falling over the last decade.
Cooling is a potential additional source of household demand but remains relatively small. NESO projects residential air-conditioning demand of only 0.5–0.8 TWh in 2035.³³ While this may shift in coming years, for now we have not included this as a major demand driver.

Third, even if these levels are reached, this is not net new demand for the economy. It is driven by switching from other fuels.

Fourth, industrial energy demand comes predominantly from chemicals, petrochemicals, glass, ceramics and fertilisers. Lower electricity prices could in theory support growth, but this heavy industry has declined consistently since the 1970s for broader reasons that predate Britain’s high electricity prices.

Data centres therefore stand out as the strongest source of near-term growth, with demand that has the potential to be almost entirely economically additive, bringing new investment and supporting AI sovereignty.

This prompts two questions. First, how much additional electricity demand could Britain feasibly add over the next decade through data centres? Second, how can this be done in a way that supports the grid rather than adding further costs and constraints?

NESO, based on its recent call for inputs, estimates the mature pipeline of data centres with committed financing at 20GW,³⁴ up from 2.4GW installed as of 2024.³⁵ Based on a variety of external forecasts, we estimate that between 9GW and 17GW is achievable by 2040.³⁶ Drawing on this, and on the fact that Virginia has been adding data centres at a rate of 1GW per year,³⁷ while Ireland grew its data centre capacity at a CAGR of 20% from 2019 to 2023,³⁸ we estimate Britain could feasibly add 14GW by 2035. This represents our upside demand scenario and would see GB demand reach 425 TWh by 2035 (Exhibit 8), reducing consumer bills by £35 in 2035.³⁹

It is also important to note that water constraints are a key consideration when it comes to delivering data centre growth at scale. Requirements vary significantly by cooling technology and location, but in some areas large facilities can place material additional pressure on local water resources, particularly where supply is already constrained. We have not assessed this risk in detail here, and any large-scale build-out will need to account for water availability alongside power, land, fibre and other local infrastructure.

How additional demand is charged will also matter. Whether additional volume helps recover the fixed costs already in the ground depends on the design of network charges. A key principle in designing these charges is to make sure that they are progressive. It should not be that those that cannot afford to adopt new technologies, such as EVs or solar, are left paying more because of those that can.

The objective is not to connect data centres at any cost, nor to assume that new demand requires no new infrastructure. It is to maximise the value of that demand to the system. That creates a simple test: bring forward demand where the value it adds exceeds the costs it creates, and make the developer, not existing consumers, bear any incremental costs. This is not the approach taken internationally so far. Britain has a chance to learn from this.⁴⁰

Faster connection should come with firmer commitment, due to the long life of network assets. Britain should draw on approaches emerging in US markets, tying large-load connections to minimum-term commitments backed by financial guarantees, typically over five to seven years, so that if the load does not materialise the developer carries the cost rather than the billpayer.

Capturing this opportunity requires solving the principal constraint on data centre growth: access to power. This is primarily a network-capacity problem rather than a shortage of supply. Renewable generation is growing fastest in Scotland and northern England, while demand is concentrated around London and the South East, where transmission capacity is scarce. Britain can have abundant generation and still make projects wait years.

Rectifying this means changing the relationship between data centres and the grid. Rather than treating each site as an inflexible base load for which the network must first be reinforced, data centres can become grid-positive, locating where power is available and adapting demand to grid conditions.

International markets are already moving in this direction. Texas’s Batch Zero process assesses large loads together, establishing where demand can connect reliably and what upgrades are needed.⁴¹ Virginia created a dedicated large-load rate class, requiring customers to pay at least 85% of the transmission and distribution costs of serving them each month. Under Ireland’s new connection policy, new data centres must provide generation or storage matching their requested capacity and source at least 80% of their annual demand from additional Irish renewables.

A recent BCG paper, Powering AI through Grid-Positive Data Centres, sets out four levers for putting this into practice. Below, we apply these to the British context.

Trade flexibility for an earlier connection. Government and Ofgem could introduce a voluntary flexible connection for users above 50 MW, allowing data centres to connect earlier in return for predefined curtailment. NESO, networks and data centre operators would negotiate limits on hours, duration, notice and capacity. That would allow operators to size the behind-the-meter assets needed to manage curtailment. Modest flexibility creates substantial headroom. Duke University found that curtailing flexible loads for up to 1% of annual hours, around 88 hours, could unlock 126 GW of additional load across 22 US balancing areas, roughly 16.6% of aggregate peak demand.⁴² That translates into faster connections, cutting time to power from 5-10 years to 2-4 years by avoiding or deferring major reinforcement. This represents a roughly 60% reduction in connection time.
Create a British market for powered land. We have identified around 18GW of potential powered land not currently in use, drawn from public information on former power-generation and brownfield sites, with around 40% showing no current redevelopment activity.⁴³ Government could task NESO, transmission owners and distribution networks with publishing a national Powered Land Register showing where large demand could connect fastest. This would combine releasable grid capacity, network constraints, land availability, connection rights, planning status and proximity to fibre. Ofgem could add a ‘use it, trade it or lose it’ mechanism to transfer unused capacity to credible projects. Government could also link support to measurable system benefits such as absorbing otherwise-curtailed renewables and avoiding expensive reinforcement.
Let developers build their own connections. Ofgem’s June 2026 Connect update proposes two self-build routes: developers could build high-voltage assets and transfer them to the Transmission Owner or retain them under an Independent Transmission Owner licence.⁴⁴ Government could scale this by standardising build-and-transfer contracts, including pre-agreed terms for valuation, warranties, liability and asset transfer. Ireland provides a precedent, with Ofgem citing projects delivered around 20% more cheaply and 11 months faster where the build was done contestably.⁴⁵ Self-build does not create shared-network capacity, but it can remove delivery bottlenecks. BCG’s paper Powering AI through Grid-Positive Data Centres estimates that earlier energisation can create more than US$8 million per megawatt in net present value for the data centre operator. Separately, at a system level, every 1GW of grid-positive capacity could create US$1–2 billion in value from better infrastructure utilisation and US$0.5–1 billion from avoided generation and grid reinforcement.
Make behind-the-meter power a regulated bridge to the grid – or a permanent alternative to grid reinforcement. Government and Ofgem could allow data centres to open before their full grid connection is available, using dependable behind-the-meter generation, storage and flexibility to meet the balance. This should not mean fully off-grid operation, which would remove any contribution to wider system costs. Instead, it supports two forms of flexible connection.

Under a temporary connection, the site contracts its ultimate firm capacity upfront but releases it in stages, filling any gap through behind-the-meter power and transferring load to the grid as reinforcements are completed. Where that generation is substantial, initial grid requirements could fall by more than 50%. West Sleekburn illustrates the potential, with developers proposing 90 MW of natural-gas fuel cells to support around 50 MW of IT load as a temporary bridge while the permanent solution is built.⁴⁶

Under a permanent alternative, the site retains lower firm capacity and uses behind-the-meter assets to manage curtailment over its lifetime, so reinforcement happens only where the wider system shows it is needed. Charging would need to reflect the distinction. Under the temporary model, developers commit from day one to the reinforcement costs attributable to their ultimate demand, with network charges rising as capacity is released, while a permanent connection needs charging that reflects enduring reliance on lower firm capacity and still contributes appropriately to shared costs. Crucially, a permanently flexible data centre would remain connected to the grid and pay network charges for the capacity and energy it uses. Its contribution to shared system costs may therefore be lower than under a fully firm connection, but the relevant alternative in constrained areas may be that the data centre cannot connect at all.⁴⁷

Build local acceptance alongside solving the power constraint to enable data centre construction at scale. With concerns around impacts on land, infrastructure and communities, rollout will need to deliver tangible benefits for host areas from the outset. This ranges from jobs and skills to local infrastructure, public services and the environment. The evidence so far suggests visible improvements to the local area are worth more than unassigned funding. A data centre academy in Phoenix has created local training and employment pathways.⁴⁸ In Odense, surplus heat from a data centre supports district heating for around 6,900 households.⁴⁹

The goal is to make data centres fit the grid Britain has today, while building the grid it needs tomorrow – thereby bringing forward the new demand that can help break Britain's negative price-demand spiral.

Weekly Insights Subscription

Stay ahead with BCG insights on energy

Methodology

This analysis is built on BCG Center for Energy Impact's GB Power Model – a bottom-up model of Britain's electricity system that reconstructs every component of the household bill separately from the underlying cost of running the system.

Forward demand assumptions are taken from the Department for Energy Security and Net Zero's (DESNZ) central forecast. Policy assumptions are estimated from historic trajectory and announced government policy. Generation capacity is based on Aurora Energy Research’s forecast, with our own assumptions layered on to reflect utilisation and dispatch behaviour.

The wholesale electricity price is generated by a merit-order dispatch simulation of the GB power system: generating plant is ranked cheapest-first by running cost, and the price in each half-hour is set by the most expensive plant needed to meet demand. Demand is dispatched against this merit order on a 30-minute basis, across a sample of representative days each year, to capture the daily demand curve and seasonal fluctuations. This simulation is calibrated against actual 2024 wholesale prices. Network charges, policy levies and supplier/retail costs are each built up separately from their own real-world data. These four components are then combined with the Contracts for Difference top-up and the Capacity Market charge, to give a pre-tax total, with VAT then applied. All of these components are considered for a typical household using 2,500 kWh/year. The Contracts for Difference top-up is recalculated on the model's own wholesale price, rather than an external forecast, so it automatically adjusts if that price moves. This means a cheaper or more expensive wholesale outcome feeds mechanically back into the bill. The Capacity Market charge is based on real capacity-market cost data, updated to reflect the most recent auction result.

All three of our scenarios run through the same model and differ only in the assumptions fed into it. The central baseline takes DESNZ's central demand forecast as given. The low-demand case keeps every other assumption unchanged but assumes demand continues falling as it has for the past decade, rather than recovering. The upside case starts from the central baseline and changes five inputs: faster growth in data centre electricity demand; some policy costs moved off bills and onto general taxation; a longer assumed network asset life; a lower cost of capital for network investments; and some network capital costs deferred (beyond the time horizon of the model) rather than recovered immediately through bills.

Network costs follow published regulatory price-control determinations, but scale with each scenario's own demand: grid investment is sized to that scenario's peak demand and recovered, alongside the existing asset base, per unit of electricity consumed – so a lower demand path spreads a similar level of fixed network costs over fewer units, resulting in a higher network charge per kWh. Each policy levy grows at its own long-run rate. Every scenario is run through this same calculation for each year to 2035, producing the separate bill trajectories used in this report.

All figures are in nominal terms (not adjusted for inflation).

About Us