Luke Sperrin
Head of Clean Energy Industries, Digital Catapult
AI-led solutions can potentially build and support the infrastructure to meet growing electricity demand. But too often, debate is dominated by concerns around AI’s electricity use.
For too long, AI, compute and advanced connectivity technologies have been thought of in isolation, each treated as a separate infrastructure challenge. They must be considered collectively to improve existing infrastructure and unlock new opportunities.
Using AI to better utilise existing infrastructure
One area where AI can make a difference is asset management and network planning. The IEA estimates that widespread AI use could unlock up to 175 GW of additional transmission capacity from existing power lines, allowing operators to better utilise existing infrastructure.1 This is equivalent to approximately three times Great Britain’s 2024 peak electricity demand.2
The UK Government’s Review of AI Deployment in the Electricity Networks also identifies forecasting, optimisation and energy flexibility as areas where existing approaches will need to evolve as the electricity system becomes more complex.
Together, these factors can help make better use of existing infrastructure
and limit the additional pressure AI places on the electricity system
Moving towards whole-system optimisation
Improving individual assets is only part of the answer. The UK’s energy, compute and connectivity infrastructure is increasingly interdependent, and planning each separately risks creating inefficiencies elsewhere. Decisions on where to locate new data centres, for example, need to consider grid capacity, access to renewable energy and storage and high-capacity connectivity.
Together, these factors can help make better use of existing infrastructure and limit the additional pressure AI places on the electricity system. DeepMind has demonstrated that machine learning can increase the value of wind energy by approximately 20%, using AI-based forecasting to better predict wind power output and optimise commitments to the electricity grid.2
At Digital Catapult, we’re working with industry to support AI integration in telecommunications networks and its responsible use across energy networks and infrastructure. This includes giving businesses the expertise and environments to test new approaches and make informed decisions about deployment.
The challenge is no longer simply proving what AI can do. That’s why we’re providing high-quality, interoperable data, trusted environments for testing new applications and closer collaboration between technology innovators and infrastructure operators to turn promising ideas into real-world infrastructure.
[1] IEA. AI for energy optimisation and innovation. tinyurl.com/5n8wsrb4.
[2] NESO, 2025. Future Energy Scenarios: Pathways to Net Zero. tinyurl.com/y7kdwvwy
[3] Elkin & Witherspoon, 2019. Machine learning can boost the value of wind energy. Google DeepMind. tinyurl.com/57h88xrx
