Net Zero and AI together represent the largest challenge to the grid in a generation writes Seb Kirk, Co-founder and CEO of GaiaLens
For the UK power sector, the next five years will be defined by two key forces: the first of these is the drive to decarbonise the grid – pushing towards Net Zero through greater use of renewables, new nuclear, electrification and smarter system balancing. The second is the rapid expansion of AI and the data centre power load that comes with it.
Each on its own is transformative. Together, they create a serious challenge for the sector as a whole: the National Grid, distributed network operators (DNOs) and large energy users. But they also point towards the same solution: delivery of a far more data-driven, dynamic and flexible electricity system.
As such, AI is not just another source of demand hitting an already over-stretched network. It is becoming one of the key ways the sector will manage rising complexity, unlock new flexibility and create new customer value.
Electricity demand rising
The broad shape of the challenge is clear enough already. Britain’s electricity system is being asked to do more, and to do it differently. As transport, heating and parts of industry electrify, overall power demand is set to rise significantly. The National Energy System Operator (NESO) has said Britain’s electricity needs could increase by up to 15% by 2030 and by 40% by 2035. At the same time, the Government’s Clean Power 2030 programme is pushing for a much faster build-out of low-carbon generation and a more flexible, digitally enabled system to manage it.
That alone would be enough to keep networks, suppliers and flexibility providers busy. However, AI is adding another layer of pressure, and a particularly awkward one. The International Energy Agency’s (IEA) April 2026 report says global electricity demand from data centres is projected to more than double by 2030, with Europe’s data centre demand rising by more than 45TWh, or 70%, over the same period. The fastest-growing component is the accelerated compute used for AI workloads.
That matters because data centre demand is not just representing greater power demand. It is concentrated demand, often in grid-constrained locations with very high reliability requirements and very little tolerance for interruption.
UK big data centre build out is underway
The UK government’s 2024 estimate put Britain’s co-location market at around 1.6GW of IT load, with London alone accounting for just over 1GW. However, the pipeline now coming into view shows the next phase of data centre building is likely to be much larger. North Lincolnshire has granted consent for a 1GW data centre campus at Elsham Tech Park. QTS is also progressing a development at Cambois in Northumberland that could eventually total 720MW across 10 buildings. Those are not marginal additions. They are major new loads entering the planning and infrastructure system.
In electricity demand terms, the UK data centre sector is still relatively modest today but not trivial. A London Assembly paper said UK data centres accounted for less than 10TWh of electricity demand in 2024, against total UK electricity demand of 319TWh. That puts the sector close to three per cent of national demand just over a year ago.
The IEA’s latest Europe-wide demand trajectory gives us a sensible working assumption that UK data centre electricity demand could rise to 11TWh this year and increase to 18TWh by 2031 – making it nearly six per cent of overall UK electricity demand by 2031.
AI is a big problem but also offers well spring of innovation
It would be easy, therefore, to frame AI simply as a new electricity ‘hog’ arriving at exactly the wrong time, just as Britain tries to decarbonise and electrify more of the economy. There is some truth in that. AI-focused data centres will add real load and, in some places, real system stress.
However, AI is also part of the solution if you look deeper into its likely uses. AI will be used by utilities and energy-tech providers to optimise energy use, improve customer economics and build a more resilient, flexible system – all characteristics that it badly needs to modernise.
That matters because the future power system cannot be run efficiently with analogue-era operating models. A grid with much higher renewable penetration, millions of electric vehicles, more batteries, more heat pumps, and materially higher digital infrastructure load needs far better forecasting, more granular demand visibility, faster decisioning, and much more automated flexibility – areas where AI systems can help.
Kraken factor
That is exactly the territory where smart data combined with AI systems start to earn their keep. Kraken is a good company to watch in this regard. The company is often described simply as a billing platform with a modern interface. However, that understates what it has become.
Octopus says Kraken is now contracted to support more than 70 million household and business accounts worldwide and processes around 15 billion new data points a day. It’s now a smart data management and usage powerhouse. It is already used by utilities including EDF, E.ON Next, National Grid in the US, Origin Energy, Plenitude and Tokyo Gas.
What makes Kraken important is not just its scale but the operating model it represents. At home, the clearest example is Intelligent Octopus Go. Kraken powers this service that automatically schedules EV charging into cheaper and greener time windows. Kraken says the tariff now covers more than 160,000 connected EVs, processes more than 100 million readings a day, and represents over 700MW of connected power available to support system balancing.
That is not just a nice retail proposition. It is a demonstration of how customer devices can be turned into flexible grid assets while also improving customer outcomes. Lower bills for households, better use of renewable generation, and less strain on the system at peak times, is exactly the kind of triple win the sector needs more of.
The same logic applies to Saving Sessions. Saving Sessions is a smart, data-driven demand-flexibility scheme which shows how utilities can use automation, granular consumption data and platform intelligence to lower customer bills and support grid resilience.
Over the winter 2023-24, more than 1.5 million participants took part in Saving Sessions, unlocking 2GWh of flexibility and enabling £5.1 million to be paid back to customers. This is important because it shows how data, automation and customer engagement can be turned into a meaningful flexibility product.
Abroad, the model is travelling well. Kraken’s flexibility partnership with EDF is designed to optimise the use of customer devices such as EVs, batteries, solar and heat pumps, shifting demand towards cheaper and cleaner periods. EDF says over 1,000 low-carbon devices are already connected in the current phase, representing around 7MW of flexible capacity.
E.ON Next has also said the move to Kraken has helped free teams to focus on new products and propositions, while Good Energy is using Kraken for commercial and industrial customers, allowing new tariffs to be built in minutes instead of weeks and reducing inbound business customer calls by 32 per cent. Kraken shows what the sector increasingly needs: not just digitisation but orchestration. Inevitably, this means that AI can help manage flows of data in and out of systems; as well as real-time analysis of pricing and consumption data for the benefit of suppliers, distributors and customers alike.
Value added services
For years, much of the energy transition conversation has been dominated by infrastructure volumes: more renewables, more battery storage, and in the near future more ‘grid edge’ reinforcement, together with more connections reform and more network investment.
All of that remains true. However, the coming AI boom makes something else equally important: the ability to coordinate a much more dynamic system in real-time. Utilities will need to get better at shifting load, forecasting customer behaviour, optimising distributed assets, matching flexible demand to renewable output, and packaging all of that into customer propositions that people actually understand and value. That is where AI-enabled smart data becomes more than an internal efficiency play. AI becomes the source of multiple value added services.
AI systems will increasingly be used to support:
- Lower system costs through better balancing
- Reduced customer energy spend through optimised timing
- New flexibility revenues for homes and businesses
- Improved economics for EVs, batteries and heat pumps
- More resilient network operation through better visibility and control.
This is also where this value-added services layer starts to power up. The winners will not just be suppliers or networks that know how to move electrons. They will be the ones that can turn data, flexibility and automation into useful valued added services and lower bills for customers.
AI as enabler of vital modernisation
AI-driven data centres will unquestionably add pressure to the grid. They will sharpen questions around connection queues, local constraints, resilience and overall power system investment. However, AI will also underpin many of the innovations they we will need to make the grid the flexible, smart system it needs to become.
AI is going to become part of the operating logic of the modern power system. Used well, it can help utilities reduce waste, shift demand, improve customer revenues, unlock flexibility and manage the very volatility that a renewables-heavy, electrified economy inevitably brings.
That is why the real opportunity is not in using AI for a few marginal productivity gains around the edges. It is in using AI, together with smart data, to extract vital insights and help build a system that is more flexible, more responsive and more commercially intelligent.
That is the direction the market needs to travel in, and if the UK is serious about navigating both the Net Zero transition, and the AI-driven ‘power boom’, without simply piling cost onto the system, it will need many more AI-powered platforms and propositions that make the grid not just bigger, but smarter.
