Can AI Fix the Grid it Depends On?
by Martha Shaylor on 19 Nov 2026
When Prime Minister Andy Burnham announced Great British Grid in his first conference speech as Labour leader in September, he put the electricity grid at the heart of British politics. The new, publicly owned branch of Great British Energy is intended to speed up grid connections and help bring the UK’s energy costs in line with its European neighbours by 2035, commitments reaffirmed in the October Budget.[i] It is a political — and hardly surprising — choice: with oil prices rising, climate pressures intensifying and new technologies driving ever more demand for electricity, the UK needs a grid capable of handling more power and more unpredictability. And at present, it is struggling to do so.
Burnham’s announcement caps a year of growing policy activity as the government tries to reconcile two priorities: reducing energy costs while sustaining progress towards Clean Power 2030 (CP2030) and wider Net Zero commitments, where affordability is increasingly taking precedence. Remediating this tension is, in part, blocked by the persisting grid bottleneck, where transmission networks simply cannot handle either the available supply or growing demand. Artificial intelligence (AI) has been proposed as a primary tool for creating a more efficient, responsive electricity grid, a focus expected to be set out officially in an AI for Clean Energy Strategy in early 2027.
And yet the paradox is that AI also adds pressure to the grid, demanding more electricity and storage than ever before. So whether AI is ultimately a help or a hindrance may depend on choices the government makes over the coming year — choices businesses can help shape.
A squeeze from both ends
The UK’s grid has been under pressure for years. On one side, connection queues and ageing transmission infrastructure limit its ability to accommodate new generation and transport electricity across the country. Much of the existing infrastructure was built between the 1950s and 1970s, when the energy system was organised around large, centralised power stations. Today, transmission networks increasingly struggle to move electricity from where it is generated, such as on wind farms in Scotland, to where it is needed elsewhere in the UK. According to the National Audit Office (NAO), wind and solar farms are being paid to switch off when the grid cannot handle and transport the amount of energy produced, while gas-fired power stations closer to centres of demand are paid to switch on, adding to costs for consumers.[ii]
This is why the UK’s progress in generating clean power does not automatically clear the path for a fully renewable future. In 2025, renewables alone supplied a record 52.5% of UK electricity generation, marking the second straight year in which they accounted for more than half of total generation,[iii] and also went a full year without coal power for the first time in more than 140 years.[iv] These are big milestones. And yet clean electricity per se is not enough when the infrastructure itself cannot conduct it as needed.
On the other side of the squeeze, technological advancements and the energy transition itself are creating ever-greater demand for electricity, putting more pressure on an already struggling grid. As transport and heating shift away from petrol, diesel and gas towards electricity, and as innovation drives demand for energy-intensive infrastructure such as data centres, the grid will need to carry significantly more power than it does today. For example, National Energy Systems Operator (NESO) projects that Britain’s electricity demand will rise by more than 30% by 2035;[v] globally, data centres are expected to account for at least a fifth of electricity demand growth in advanced economies through to the end of the decade.[vi]
This growing energy footprint is already generating strong public opposition, in the UK and around the world, with thousands of campaigners voicing protest against major development projects.[vii] Tensions between the need for, as well as the energy consumption and local impact of, data centres were reflected in the new Data Centre Covenant, announced in the October Budget. This proposed quicker access to the grid (the current queue for access is ten years!) for data centre projects that meet new targets for limiting water pollution, creating new jobs, and working with local communities, as well as for allowing more temporary onsite generation. The Covenant, however, only addresses part of the problem, as the UK still needs a grid capable of handling these changes quickly.
AI as the fast lever
Lucy Yu’s independent review The Grid We Need Now, published in September 2026, argued that the grid’s problem was more about rules than wires.[viii] Deterministic security standards insure against pre-specified worst cases, regardless of actual conditions — such that, wind generation, for instance, is curtailed not due to capacity limitations, but to hedge against exceedingly rare contingencies. Yu suggests AI can help reduce this need for headroom with better forecasting and modelling, allowing a shift to a more probabilistic, risk-based and, crucially, efficient system.
This proposition was echoed in the government’s own Vision for an AI-Enabled Clean Energy System call for evidence, which made the case for using existing infrastructure more efficiently with AI, such as by predicting demand more accurately or spotting faults before they happen, ultimately making energy more affordable.[ix] There is already progress being made, with the National Grid using dynamic ratings on 275km of transmission lines, estimated to save around £20 million a year,[x] and NESO aiming to rollout a Dynamic Reserve Setting in early 2027 — a tool using probabilistic machine learning to decide how much reserve power to hold, and expected to save the UK another £30 million per year.[xi]
Will AI help or hurt the grid?
Whether AI ends up a net help or net burden depends on three conditions, the rules for which are still being written.
The first is that the right kind of AI be used, and that the energy consumed by an AI-powered grid be clearly accounted for. The oxymoronic premise ‘AI for clean energy’ is well established — since AI, as a tool to reduce emissions and streamline renewables, is itself hungry for energy and water. But studies are beginning to weigh AI’s environmental cost against the emissions an AI-run grid could avoid. For example, LSE’s Grantham Research Institute estimated in 2025 that AI applications in just three sectors (power, food and mobility) could cut global emissions by 3.2 to 5.4 GtCO2e per year by 2035 — more than offsetting the projected 0.4 to 1.6 GtCO2e increase from all AI-related data centre activity.[xii] A related distinction is drawn by techUK: generative AI uses much more energy than more task-specific software; the smaller, domain-specific models best suited to running the energy system are much less energy consuming.[xiii] The question, therefore, is not whether AI uses energy — but which kind of AI, for what? How the government navigates these various choices will become clearer in its forthcoming AI for Clean Energy Strategy.
The second condition is clear accountability. The government’s Vision acknowledges outcomes from thousands of small automated decisions are not easily traced back to any single choice or actor. If an AI-optimised grid fails, who is responsible: the operator, developer or approving body? Yu’s review also recognises this gap and recommends that government clearly define approval authorities for AI-enabled autonomy and develop a long-term roadmap for updating accountability and liability rules. Until such frameworks exist, operators will likely remain cautious — and critical AI tools undeployed.
The third condition is the capacity to implement needed changes. Skills, supply chains, procurement and legacy IT are all potential bottlenecks in themselves. The NAO has flagged skills shortages and supply chain constraints as barriers to upgrading the grid, and Yu’s review found that AI tools often stall on slow procurement and outdated IT systems. Even the best-designed AI will deliver little if it cannot be bought, integrated and scaled.
All three conditions are still being shaped through live consultations, Ofgem’s AI assurance work and the 2027 strategy — stakeholders who engage now can influence outcomes.
Opportunities for business?
Writing in the New Stateman in 2023, National Grid’s Carl Trowell described the energy transition as the biggest economic opportunity of the 21st century.[xiv] Here are four opportunities for businesses as we near the end of 2026:
Grid-optimisation and AI firms — Yu’s review recommends open forecasting benchmarks by the first half of 2027; a possible forecasting marketplace to circumvent slow procurement; and a national AI for Grid Innovation Delivery Hub. These proposals could create new routes to market for businesses developing software for electricity forecasting, network optimisation and grid management.
-
-
The forthcoming AI for Clean Energy Strategy will clarify which recommendations the government intends to take forward: businesses should consider how their products could address the grid’s operational challenges and prepare to engage with emerging testing and procurement arrangements.
-
-
-
Developers should therefore assess whether their projects meet the Covenant’s requirements and consider responding to the draft Data Centre National Policy Statement consultation by 8 January 2027.[xvi]
-
AI assurance and compliance — As AI becomes more involved in electricity network planning and operations, energy companies will need to demonstrate that these systems are safe and properly governed. Ofgem’s AI technical sandbox will provide a controlled environment for companies to test their AI application, and Yu has called for clearer rules governing increasingly autonomous grid operations.
-
-
Energy companies and technology providers should therefore consider engaging with Ofgem’s AI Reg Lab and the emerging assurance guidance to help shape expectations and prepare products for adoption.
-
Infrastructure and engineering suppliers — The NAO has highlighted supply chain constraints and skills shortages that could delay the £70bn of grid upgrades needed.[xvii]
-
-
This creates opportunities for businesses supplying network equipment, specialist engineering skills and digital integration services. Suppliers should therefore engage early with network operators and infrastructure contractors to position themselves for upcoming projects.
-
Looking ahead, the end of 2026 and start of 2027 should bring greater clarity on the UK’s approach to AI in energy. This will be complex and evolving, given the speed at which AI is developing, therefore stakeholders will need to keep pace with a highly dynamic policy landscape.
If you would like to consult on anything discussed in this blog or would like an introductory chat to find out what Inline can do for you, please reach out at: enquires@inlinepolicy.com.
[i] HM Treasury (2026) ‘Budget 2026’. Available at: https://assets.publishing.service.gov.uk/media/6ae20b85460c1346f7be784e/E03668986_HMT_Budget_2026_ELAY__2_.pdf
[ii] National Audit Office (2026) ‘Upgrading the electricity transmission network’. Department for Energy Security & Net Zero. Available at: https://www.nao.org.uk/wp-content/uploads/2026/09/Upgrading-the-electricity-transmission-network.pdf
[iii] RenewableUK (2026). Available at: https://www.renewableuk.com/news-and-resources/press-releases/record-breaking-stats-show-renewables-generated-over-half-uk-s-electricity-for-the-second-year-running/
[iv] Electricity Live (2026). Available at: https://electricitylive.com/records/coal-free
[v] National Energy System Operator (NESO) (2026) ‘Beyond 2030 Update’. Available at: https://www.neso.energy/document/383376/download
[vi] International Energy Agency (2025). ‘Energy and AI Report’. Available at: https://www.iea.org/reports/energy-and-ai
[vii] The Guardian (2026) ‘Activists across Europe prepare two days of protests against datacentres’. Available at: https://www.theguardian.com/world/2026/nov/05/activists-across-europe-protests-ai-against-datacentres
[viii] Yu, L. (2026) ‘The Grid We Need Now - Independent Review of AI Deployment in the Electricity Networks’. UK Government. Available at: https://assets.publishing.service.gov.uk/media/6a9fea93c5796a7a179c641f/the-grid-we-need-now-independent-review.pdf
[ix] Department for Energy and Net Zero (2026) ‘Vision for an AI-enabled clean energy system – call for evidence’. HM Government. Available at: https://www.gov.uk/government/calls-for-evidence/vision-for-an-ai-enabled-clean-energy-system
[x] National Grid (2025) ‘National Grid installs grid-enhancing technology allowing more renewable power to flow through existing transmission lines’. Available at: https://www.nationalgrid.com/media-centre/press-releases/national-grid-installs-grid-enhancing-technology
[xi] Yu, L. (2026) ‘The Grid We Need Now - Independent Review of AI Deployment in the Electricity Networks’. UK Government. Available at: https://assets.publishing.service.gov.uk/media/6a9fea93c5796a7a179c641f/the-grid-we-need-now-independent-review.pdf
[xii] Stern et al. (2025) ‘Green and intelligent: the role of AI in the climate transition’. Available at: https://www.nature.com/articles/s44168-025-00252-3
[xiii] techUK (2026) ‘How can the energy sector leverage AI to deliver consumer benefits?’. Available at: https://www.techuk.org/insights/reports/how-can-the-energy-sector-leverage-ai-to-deliver-consumer-benefits
[xiv] Trowell (2023) ‘Why the UK needs a major energy grid upgrade now’. Available at: https://www.newstatesman.com/spotlight/sustainability/energy/2023/12/national-grid-electricity-renewables-infrastructure-wind-net-zero
[xv] Ofgem (2026) ‘Ofgem acts to free up grid capacity by tackling speculative data centre projects’. Available at: https://www.ofgem.gov.uk/press-release/ofgem-acts-free-grid-capacity-tackling-speculative-data-centre-projects
[xvi] Department of Business, Innovation, Science and Trade (2026) ‘Draft National Policy Statement for Data Centres – Consultation’. Available at: https://www.gov.uk/government/publications/draft-national-policy-statement-for-data-centres
[xvii] National Audit Office (2026) ‘Delays to electricity grid updates risk higher consumer bills’. Available at: https://www.nao.org.uk/press-releases/delays-to-electricity-grid-upgrades-risk-higher-bills/
Topics: UK politics, UK business, Regulation, GBGrid, CleanEnergy, DataCentres




Comments