An artificial intelligence (AI) tool will be used as a \u201csmart mediator\u201d between a data centre and Great Britain\u2019s electricity transmission network to manage electricity demand as part of a demonstration project led by National Grid.<\/p>\n
The project will see a live test of a tool built by a company called Emerald AI, utilising Nvidia GPUs (graphics processing units), to \u201cdynamically adjust energy consumption\u201d of a data centre in late 2025.<\/p>\n
Nvidia GPUs are key pieces of hardware used as the foundation for AI tools and have led to Nvidia becoming the most valuable publicly listed company in the USA.<\/p>\n
Trial should make better use of existing generation capacity \u2013 National Grid<\/p>\n
National Grid did not name the data centre which would be used in the experiment, but said: \u201cIncreased flexibility of data centre consumption can unlock capacity and accelerate connections to the electricity network.\u201d<\/p>\n
Explaining the rationale for the test, National Grid said the Emerald AI tool would get data centres to temporarily use less electricity when general demand for electricity from the transmission grid is high.<\/p>\n
\u201cThe electricity transmission network is designed with built-in redundancy to deliver world-class reliability and keep the lights on for customers,\u201d it said.<\/p>\n
\u201cCapacity is typically available outside of peak events like hot summer days or cold winter storms, when there\u2019s high demand for cooling and heating. That means, in many cases, there\u2019s room on the existing grid to connect new data centres, if they can temporarily dial down energy usage during periods of peak demand.<\/p>\n
\u201cThis strategic partnership will help position the UK as a global leader in AI and reduce the need for additional infrastructure by giving National Grid more confidence that data centres can lower their energy use when required.<\/p>\n
\u201cBy increasing the utilisation of existing electricity infrastructure, National Grid, working closely with the National Energy System Operator, can better manage growing demand, attract investment in advanced computing, and support the UK\u2019s transition to a more efficient and flexible energy system.\u201d<\/p>\n
National Grid said the demonstration would showcase how \u201ca wide variety of AI workload types can be adjusted in real time\u201d.<\/p>\n
The company added: \u201cBy changing computing activity when the grid is under pressure, the demonstration will prove that AI data centres can act as responsive partners to the electricity network while maintaining performance standards for mission-critical workloads.\u201d<\/p>\n
National Grid also said it \u201cmade a strategic investment in Emerald AI\u201d but did not disclose the value of the investment.<\/p>\n
The partnership will \u201cadvance technical standards for AI data centre flexibility, collaborate with customers to bring this capability to the growing pipeline of UK data centres, and foster broader industry collaboration through the NextGrid Alliance\u201d, according to National Grid.<\/p>\n
\u201cThrough this collaboration, National Grid and Emerald AI are establishing a new blueprint for integrating AI and energy infrastructure \u2013 one that shows how data centres can help make the power grid more reliable, lower costs, and support clean energy,\u201d it concluded.<\/p>\n
Data centres already power-hungry, AI data centres even more so<\/p>\n
Murphy Geospatial head of department \u2013 subsurface engineering David Sayad wrote for NCE in November 2024 where he said data centres are \u201cenergy-hungry sites, full of servers with strict security, need an uninterrupted power supply to operate and substations across the national grid will need more capacity to serve this infrastructure to make it viable\u201d<\/a>.<\/p>\n \u201cPreviously, UK data centres have been built around London. As data consumption grows, developers have started looking further afield after struggling to find adequate land and power in this area,\u201d he added.<\/p>\n The power demand of AI data centres is expected to be so high compared to the electricity available from the electricity transmission grid that AI data centre developers are looking to power them with dedicated nuclear power plants<\/a>.<\/p>\n