Written submission from Beyond Fossil Fuels, Foxglove, Friends of the Earth, Global Action Plan, Global Justice Now, Green Web Foundation and The Citizens (DCU0070)
Joint NGO response to Environmental Audit Committee inquiry into the risks and opportunities to the sustainability of data centres in the UK
We are civil society organisations working on technology and the environment, who share concerns that the un-checked roll-out of data centres is having, and will have, very significant energy use and environmental impacts that are not being sufficiently accounted for.
Summary
Energy
The new generation of ‘hyperscale’ data centres planned across the UK are on a different scale to the facilities in place today. Smaller data centres typically consume 5-10 megawatts (MW) of power, whereas the data centre recently voted for by the council in Elsham, Lincolnshire, or that planned for Cambois, Northumberland, will each have a ‘capacity’ (or demand) of one gigawatt (GW) of electricity – 200 times the energy requirements of the previous generation and equivalent to the total output of a gas or nuclear power station.
Ofgem’s recent ‘call for input’ regarding demand connection reform identified connection requests from
“around 140 (50 GW) data centres, the majority of which are likely to receive a Gate 2 offer. 71 (around 20GW) data centres reported they have achieved financial commitment with Final Investment Decision (FID), indicating a significant volume of mature data centre projects in the demand queue”[1]
For context, peak electricity demand for the entire UK during winter 2025/26 was 45GW.[2]
This additional demand - unless meaningfully matched with new, renewable energy generation and storage - poses an enormous threat to efforts to decarbonise the UK’s electricity grid.
Emissions
Following recent scrutiny, the Government has withdrawn its estimate of a ‘maximum of 0.142m tonnes of CO2 (MtCO2) from 11.2GW of AI-related computing power by 2035’. Recent analysis confirms just how significantly this was an underestimate - a more credible figure, assuming just the 20 GW of ‘mature’ projects identified by Ofgem go on to be built, suggests associated carbon emissions will be ‘as high as 70MtCO2, the annual emissions of Sweden’. This is 500 times the previous Government estimate.[3]
Even facilities smaller than the 1GW giants will, according to their own developers, contribute hundreds of thousands of tonnes of CO2 to the atmosphere every year. Google’s planned data centre in Essex will, according to the firm’s planning documents, “lead to a net increase in GHG emissions of 568,727 tonnes CO₂е per year during the operational phase.” [4]
The above estimates assume data centre energy demand will be met by grid electricity, the generation mix for which will vary according to time of use and the rate at which renewable energy is built.
But developers are now looking to build on-site gas generation to meet their power needs and, presumably, to facilitate faster build-times when grid connections are not available. One developer quoted in Utility Week noted:
“Data centres can’t connect to the electricity network today… So they come to the gas network. They’re coming to us and asking for a connection to the gas network and saying they’ll build a small gas power station to power locally their data centre. And because we’ve got the capacity, we can just do it.”[5]
The first project to have been directed into the Government’s Nationally Significant Infrastructure Project (NSIP) planning regime is a proposed 300 MW data centre in Wapseys Wood, Buckinghamshire, that is not seeking a grid connection and will instead be powered entirely by on-site gas turbines ‘at least in the initial phases’.[6] Using the Government’s own GHG conversion factors, we estimate emissions from this one data centre will be more than half a million tonnes – 532,696 tCO2e – annually.[7]
Burning such enormous quantities of gas on-site will also inevitably have a negative impact on local air quality, something that will be of concern to the two nearby primary schools and local hospital.
The emergence of gas fuelled on-site generation is particularly worrying given the absence of any National Policy Statement (NPS) that should be guiding the decision-making process of projects directed into the NSIP regime, and which should be considering carbon budgets in the round. Matthew Pennycook, Minister for Housing and Planning, said the NPS would be published “shortly” after regulations were laid in January, but to date this has not emerged.[8]
Water
Facilities using water to cool their computing equipment may cause a significant drain on water supply which could have knock-on effects on the local and national environment, straining ecosystems and intensifying competition for resources between agriculture, communities and businesses.
There is often a lack of transparency from developers on these impacts, and the Environment Agency has said that they “are experiencing barriers in gaining information about water consumption” by data centres, and “more transparency is needed”.[9]
Data centres that use water for cooling draw down an enormous amount to manage the heat generated by densely packed servers. Cooling towers, chillers, and humidification systems are the main direct uses, while large volumes of water are also consumed indirectly through electricity production to power the data centres. A mid-sized 15 MW data centre may consume around 300–500 million litres of water each year - roughly equivalent to the annual domestic water use of 2,000–3,500 UK households - while a single hyperscale data-centre campus can draw around 1.5 million litres per day, comparable to the daily water needs of a town of 10,000 people, placing it among the largest industrial water users in many local catchments.[10] Roughly a quarter of this footprint is direct onsite use.[11] Much of this is water used in cooling systems, frequently sourced from potable (drinking-quality) supplies. Some water companies have objected to data centre plans due to the use of potable water.[12]
Thames Water, the largest water company in the country, has given even higher estimates, stating in its 2025-30 business plan that “a large facility might
use anywhere between four and 19 million litres of water per day.”[13]
While alternative technologies—such as direct-to-chip or immersion liquid cooling, closed-loop systems, and air economisation—can reduce dependence on potable water, their uptake remains uneven.[14]
Water consumption by data centres is expected to rise sharply, driven in particular by the rapid expansion of artificial intelligence and machine-learning workloads. Globally, water withdrawals are projected to reach between 4.2 and 6.6 billion m³ by 2027, with onsite use in the United States potentially doubling or quadrupling by the end of the decade[15]. Company disclosures already indicate multi-billion-litre annual consumption with sustained year-on-year growth[16].
While comparable national projections are not yet published for the UK, the implications are clear. Even a modest share of this global growth would translate into hundreds of millions of additional litres of annual demand concentrated in a small number of English catchments, many of which are already classified as water-stressed. This trajectory coincides with intensifying scarcity: nearly three quarters of the world’s population currently lives in areas classified as water-insecure or critically water-insecure[17], while large parts of England - particularly the South East and East - are already experiencing structural supply deficits.
The clustering of hyperscale data centres in regions with fragile water availability therefore raises acute risks of local competition with households, agriculture and other industrial users. While efficiency improvements in cooling and infrastructure are often cited as mitigation, multiple studies caution that such gains are likely to be outpaced by the growth in computational demand, resulting in a net increase in water withdrawals rather than a reduction[18].
There is also a trade-off between the energy and water demands of data centres: to save on water, data centres use more energy-intensive cooling systems, putting more strain on the grid, increasing potential carbon emissions and contributing to indirect water consumption for power generation.[19] Giving evidence to the Environment and Climate Change Committee in January 2026, global product design lead for NTT Global Data Centers and water working group chair for the Climate Neutral Data Centre Pact Steven Campbell-Ferguson explained that many developers choose to use evaporative systems when water can be secured from the local network to save on electricity and improve the data centre’s economic business case, with the cost of energy 10 to 15 times that of water[20]. He added that [21] use of evaporative and non-evaporative systems in the UK has changed over time:
More than 50% of data centres in the UK are operating non-evaporative cooling systems. Over the last 20 years, systems have typically been designed probably about 50:50. Twenty years ago, everything was dry because it was the cheapest way. The original dotcom boom 25 years ago was dry systems, because it is lowest capex and people, frankly, did not care about energy efficiency. The industry then got very interested in energy efficiency and water-using systems. Now, the pendulum has swung back, in that we cannot get the water, so we are going back to dry systems.
Problematic water usage claims from the industry focus only on direct on-site use, entirely omitting the vast volumes of water consumed indirectly through electricity generation. International research shows that training a single cutting-edge AI model can evaporate hundreds of thousands of litres of clean water, and that global AI-related water withdrawals could reach 4.2–6.6 billion m³ per year by 2027[22]. This is higher than the household usage of the whole of the UK[23]. Yet in the UK, data-centre operators are not required to disclose their water use, leaving regulators unable to assess cumulative impacts or plan for the future.
Policy responses have so far focused on demand management - compulsory metering in “seriously water-stressed” regions, efficiency programmes and public engagement campaigns - while leaving structural reform largely untouched. Current drought plans from water companies have been criticised for over-relying on behavioural change and for inconsistent public engagement[24],[25]. Experts increasingly call for integrated water resources planning that recognises the interdependence of land use, climate volatility and infrastructure capacity[26],[27].
Recent economic modelling underscores the scale of the challenge: without major new investment, demand in England starts to exceed supply by 2034[28], and combined industrial, residential and digital infrastructure demand could produce a public water supply shortfall of around five billion litres per day by 2055 - more than one-third of current daily usage[29]. Crucially, the same modelling identifies rapidly expanding sectors such as data centres and AI infrastructure as under-recognised drivers of demand in already stressed regions. Yet the Environment Agency cannot currently model these impacts accurately because data centres are not required to disclose their water withdrawals or consumption. This lack of visibility means the UK is planning its water future with a major blind spot.
Against an already stressed system - one that must balance the needs of households, agriculture, industry and the environment - the arrival of a new, highly concentrated, fast-growing water user in the form of hyperscale AI campuses risks pushing several regions past their limits. It is within this tightening political-hydrological context that the escalating water demand of UK data-centre infrastructure must be understood.[30]
Planning policy – lack of National Policy Statement
With an estimated 100-200 proposed new data centres in the planning system already, and 140 seeking grid connections, it is crucial that the NPS is published as a matter of priority, and that it fully acknowledges and addresses the above challenges, to ensure that the public and the climate do not end up footing the environmental bill for these facilities.
At a minimum, the NPS must include:
Note: these requirements could be applied only to the developers of data centres above a certain size - for example, a capacity of 500kW - 1MW, which is above the comparable reporting benchmarks used in Germany and the EU.
Signed:
Global Action Plan
Friends of the Earth
Foxglove
Global Justice Now
Green Web Foundation
The Citizens
Beyond Fossil Fuels
[1] https://www.ofgem.gov.uk/sites/default/files/2026-02/2026-02-12-Demand-Connections-Call-for-Input.pdf
[2] https://www.neso.energy/document/379521/download The highest demand was on 5 January, peaking at 45 GW
[3] https://www.carbonbrief.org/analysis-co2-from-uk-data-centres-could-be-hundreds-of-times-higher-than-thought/
[4] See ‘Google’s huge new Essex datacentre to emit 570,000 tonnes of CO2 a year,’ The Guardian, 15/09/2025: https://www.theguardian.com/technology/2025/sep/15/google-datacentre-kent-co2-thurrock-uk-ai
[5] Jane Grey, “Electricity constraints forcing data centres to turn to gas grid”, Utility Week, 24 April 2025. https://utilityweek.co.uk/electricity-constraints-forcing-data-centres-to-turn-to-gas-grid/
[6] https://www.gov.uk/government/publications/data-centre-campus-wapseys-wood-buckinghamshire-section-35-direction-planning-act-2008 and in particular p.25, para 4.15 here https://assets.publishing.service.gov.uk/media/69b7ece3ba47c264e6c8cfba/Request_Document_-_SDC_M40_Campus_-_Section_35_Direction.pdf
[7] 300MW over a year is 2,628,000,000 kWh. 2,628,000,000 * 0.2027 = 532,695,600 kgCO2e, or 532,696 tCO2e.
[8] https://hansard.parliament.uk/commons/2025-11-12/debates/238160f2-d451-4d94-a0f8-b755e63f1a6c/DraftInfrastructurePlanning(BusinessOrCommercialProjects)(Amendment)Regulations2025
[9] See ‘National Framework for Water Resources 2025,’ Environment Agency, 9.4: ‘Water for data centres and artificial intelligence’: https://www.gov.uk/government/publications/national-framework-for-water-resources-2025-water-for-growth-nature-and-a-resilient-future/9-taking-action-on-other-significant-water-using-sectors-and-emerging-demands-national-framework-for-water-resources-2025
[10] Mytton, D., 2023. “Overestimating AI’s water footprint.” /dev/sustainability (Substack), Available from: https://www.devsustainability.com/p/ overestimating-ais-water-footprint
[11] Hiremath, R.B., 2024, November. AI-Embedded Data Centres: Promoting Sustainability and Reducing Water Footprint. In 2024 First International Conference on Data, Computation and Communication (ICDCC) (pp. 40-44). IEEE.
[12] https://www.bbc.co.uk/news/articles/crevjegnnd0o
[13] See Thames Water Business Plan 2025-30, p43: https://www.thameswater.co.uk/media-library/home/about-us/regulation/our-five-year-plan/pr24-2023/our-business-plan.pdf
[14] Bansode, S.S., Hiremath, R. and Hiremath, G.R., 2024. Promoting sustainability: Mitigating the water footprint in AI-embedded data centres. In Quality of Life and Climate Change: Impacts, Sustainable Adaptation, and Social-Ecological Resilience (pp. 220-232). IGI Global Scientific Publishing.
[15] Soares, I.V., Yarime, M. and Klemun, M., 2024. Balancing the trade-off between data center development and its environmental impacts: A comparative analysis of Data Center Policymaking in Singapore, Netherlands, Ireland, Germany, USA, and the UK. Environmental Science & Policy, 157, p.103769.
[16] Iman, N., 2025. The Hidden Costs of Intelligence: Artificial intelligence and Machine learning Adoption and the Paradox of Exponential Digital Growth. Sustainability and Climate Change, 18(3), pp.225-241.
[17] Madani K.,2026. Global Water Bankruptcy: Living Beyond Our Hydrological Means in the Post-Crisis Era, United Nations University Institute for Water, Environment and Health (UNU-INWEH), Richmond Hill, Ontario, Canada, doi: 10.53328/INR26KAM001
[18] Iman, N., 2025. The Hidden Costs of Intelligence: Artificial intelligence and Machine learning Adoption and the Paradox of Exponential Digital Growth. Sustainability and Climate Change, 18(3), pp.225-241.
[19] Environment and Climate Change Committee (2026) Corrected oral evidence: Drought preparedness [accessed 9 March 2026] https://committees.parliament.uk/oralevidence/17011/html/
[20] Environment and Climate Change Committee (2026) Corrected oral evidence: Drought preparedness [accessed 9 March 2026] https://committees.parliament.uk/oralevidence/17011/html/
[22] Li, P., Yang, J., Islam, M., & Ren, S. (2025). Making AI Less "Thirsty": Uncovering and Addressing the Secret Water Footprint of AI Models. Available at: https://arxiv.org/pdf/2304.03271
[23] Based on 2024 population estimates and average water use per capita per day for England (Ofwat (2024). Water company performance report 2023-24), Wales (Ofwat (2024). Water company performance report 2023-24), Scotland (Scottish Water (2025). Customers Save 60 Million Litres of Water a Day, But More Help is Needed) and Northern Ireland (Northern Ireland Water (2024). Why save water at home? Audit).
[24] Cook, C. (2016). Drought planning as a proxy for water security in England. Current opinion in environmental sustainability, 21, 65-69.
[25] Dessai, S., & Sims, C. (2010). Public perception of drought and climate change in southeast England. Environmental Hazards, 9(4), 340–357. https://doi.org/10.3763/ehaz.2010.0037
[26] Weatherhead, E. K., & Howden, N. J. K. (2009). The relationship between land use and surface water resources in the UK. Land use policy, 26, S243-S250.
[27] von Lany, P.H., Choudhury, F., Hepworth, N. et al. (2013). Applying Optimisation and Uncertainty Analysis to Help Develop an Integrated Water Resources Plan for South East England. Water Resour Manage 27, 1111–1122. https://doi.org/10.1007/s11269-012-0121-2
[28] National Audit Office (2020). Water supply and demand management. Available at: https://www.nao.org.uk/wp-content/uploads/2020/03/Water-supply-and-demand-management-Summary.pdf
[29] Environment Agency (2025). The National Framework for Water Resources 2025. Available at: https://assets.publishing.service.gov.uk/media/685d4a2ac2633bd820a92a99/2025_EA_National_Framework_Water_Resources_-_summary_document.pdf
[30] Not a drop to drink: How Britain’s data centre surge threatens water security, Global Action Plan, April 2026 https://www.globalactionplan.org.uk/files/not_a_drop_to_drink_-_april_2026.pdf