Rising Load, Rising Risk: What AI Is Actually Doing to Your Energy Costs

You may not have noticed it on a single invoice. But across the last two years, something has shifted in the electricity demand profile of almost every commercially active business in the UK.

It is not driven by obvious behaviour change. No new factory floor. No additional shift pattern. No dramatic expansion in floor space or headcount.

What has changed is the intensity of the systems running in the background.

Artificial intelligence is not a future consideration for energy strategy. It is already increasing load — quietly, continuously, and without clear attribution on most energy bills. For commercial property developers and asset owners, this is not a technology story. It is a structural cost story, and the data now makes that unambiguous.

The background load most businesses are not measuring

Every AI-integrated platform your business runs — whether that is a CRM, a building management system, a security layer, or a cloud-based workflow tool — draws processing power that did not exist five years ago. That processing does not happen locally. It happens in data centres running 24 hours a day, seven days a week, drawing electricity at scale.

Britain’s electricity demand rose by 1.7% in 2025, with data centres already consuming 5% of national demand in the United States and 20% of Ireland’s electricity — a pattern that prompted Ireland to place a three-year ban on new data centres to protect consumers. The UK is not Ireland, but the trajectory is the same and the pressure on infrastructure is building at pace. (Electric Insights, February 2026)

Oxford Economics forecasts that total electricity demand from UK data centres could grow fivefold over the next five years. That is not a rounding error. That is a structural shift in what the national grid is being asked to absorb — and it lands on a system that was not designed with that growth in mind.

For developers with multiple assets, each running interconnected digital systems, the aggregate effect is material. Background demand accumulates. It does not appear as a single identifiable line on an invoice. It compounds over time, and it is almost always underestimated in energy strategies built on historical consumption.

What the grid cannot quietly absorb

Rising demand does not land on a grid with spare capacity waiting to be deployed. It lands on a system already under considerable strain.

Before the recent connections reform process, the UK grid connection queue had grown tenfold in five years, reaching more than 700GW of generation and storage projects waiting for grid access — approximately four times what Great Britain requires to deliver secure, resilient, and affordable energy. NESO’s December 2025 reforms have reorganised that queue, but have not dissolved the underlying constraint. Gate 2 Phase 1 connection offers — for projects targeting connection before 2030 — are not due to be issued until between mid-May and mid-September 2026, with distribution-level offers following two months later. These are not administrative delays. They reflect the physical reality of a network being asked to carry more than it was built for. (NESO, December 2025; Solar Power Portal, February 2026)

The 140 proposed data centre schemes currently seeking grid connections are asking for a combined 50GW of electricity — 5GW more than the country’s current peak demand. That figure alone illustrates the scale of competition for grid access that commercial developers are now operating within. Every megawatt sought by a hyperscaler is a megawatt competing with the infrastructure projects your assets depend on. (Engineering and Technology Magazine, March 2026)

For developers planning solar, storage, or any on-site generation, the grid position is not a secondary consideration. It is the primary one. A well-modelled project with a poor connection position does not perform. That is not a theoretical risk. It is what we consistently see in practice.

What we consistently see across commercial projects

Across our work with developers, asset owners, and C&I clients, the same pattern appears with regularity.

Clients arrive with energy strategies built on consumption data that does not account for future operational behaviour. They have sized solar systems based on current demand. They have modelled returns against today’s load profile. They have assumed grid availability will remain stable.

When we run independent modelling that factors in AI-driven load growth, automation expansion, and increasing digital dependency, the numbers shift. Peak demand increases. The consistency of demand across the day increases. Exposure to grid pricing becomes more significant.

That gap — between what the historical data shows and what the operational trajectory actually requires — is where poorly structured energy projects fail. Not dramatically. Not immediately. But over the asset lifetime, in the form of underperformance, mispriced risk, and energy costs that were never properly planned for.

Solar and C&I battery storage are not the answer to every version of this problem. But where they are appropriately sized, correctly positioned relative to the grid, and modelled against realistic future demand rather than historical consumption, they provide genuine cost stability and a degree of independence from wholesale pricing volatility.

The distinction matters. A system designed for yesterday’s load profile will not perform as expected against tomorrow’s.

Where the commercial logic holds — and where it does not

There is a tendency in this market to reach for technology solutions before the fundamentals have been established. Battery storage revenue stack modelling, private wire arrangements, and on-site generation all have legitimate commercial application. None of them override poor grid position.

As one senior developer noted at Solar Media’s recent Solar Finance and Investment Summit: “You cannot spend blindly. Once you receive a connection date, you need to reassess the economics and make sure the project still works.” That observation applies equally to demand-side projects. A solar asset that cannot connect at the capacity modelled does not deliver the projected return. A storage system that cannot discharge when the revenue stack requires it does not perform as sold. (Spirit Energy / Solar Media, February 2026)

The starting point for any commercial energy decision is not which technology to deploy. It is whether the grid position supports deployment at all, what the realistic connection timeline looks like, and whether the load profile the system is being designed for reflects actual operational trajectory.

That sequence — grid first, technology second — is not a preference. It is what distinguishes projects that perform from projects that look credible on paper and disappoint in practice.

The global pattern the UK is expressing locally

This is not a UK-specific problem, and the scale of what is coming is not speculative.

The International Energy Agency projects that electricity consumption from AI-optimised data centres will grow by 30% annually, with global data centre electricity demand more than doubling by 2030 — reaching the equivalent of Japan’s entire current electricity consumption. In advanced economies, data centres are projected to drive more than 20% of total electricity demand growth between now and 2030, putting power sectors back onto a growth footing after years of stagnating demand. (IEA Energy and AI Report, 2025)

The UK sits squarely within that pattern. Contracted demand capacity in the UK jumped from 41GW in November 2024 to 125GW by June 2025, with data centre projects driving a significant share of that increase — a rise that NESO itself described as exceeding even the most ambitious forecasts. (Knight Frank Connections Reform Update, December 2025)

These figures do not exist in isolation from commercial property strategy. They represent the infrastructure context within which every energy decision your assets make over the next decade will play out. Understanding that context — specifically, at the level of a given asset and a given grid connection point — is what separates sound commercial decisions from ones carrying unpriced risk.

The questions that actually matter

The right starting point for any developer reviewing their energy position is not which solar technology to consider or which storage provider to approach.

It is load.

What is the actual electricity demand of this asset today — including background processes, always-on systems, and AI-integrated platforms that do not appear as obvious consumption? What does that demand look like in three to five years, given the direction of digital integration across commercial property? How exposed is this asset to grid pricing and availability constraints? And what level of genuine control over energy supply is achievable — and at what cost?

These are not abstract questions. They are the commercial foundation that every investment in on-site generation, storage, or energy infrastructure either rests on or fails to account for. Without answers to them, any energy strategy — however well-intentioned — is built on an assumption rather than an assessment.

At Independent Solar Consultants, we work with developers, asset owners, and commercial clients to interrogate energy projects against real-world constraints before capital is committed. We hold no vendor relationships and represent no manufacturers. Our position is straightforward: a project either works technically, commercially, and structurally — or it does not. Our role is to establish which is true.

In a market where digital growth is increasing energy demand faster than most businesses recognise, and where grid constraints continue to shape what is actually deliverable, independent analysis is not a discretionary service. It is the difference between a project grounded in reality and one that performs only in a model.

If you are reviewing an energy project and want it stress-tested before commitment, that is what we do.

Sources

1. Electric Insights — AI and data centres push up Britain’s electricity demand — https://reports.electricinsights.co.uk/?p=2645

2. Oxford Economics — The rising challenge of powering data centres — https://www.oxfordeconomics.com/resource/the-rising-challenge-of-powering-data-centres/

3. NESO — Connections Reform — https://www.neso.energy/industry-information/connections-reform

4. Solar Power Portal — Grid connection offers delayed again — https://www.solarpowerportal.co.uk/solar-planning/grid-connection-offers-delayed-again-as-impact-on-uk-developers-grows

5. IEA — Energy demand from AI — https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai