Written submission from Dr. Yi He (DCU0042)

Water risks and opportunities for the sustainability of UK data centres

Written evidence from Dr Yi He, Tyndall Centre for Climate Change Research, School of Environmental Sciences, University of East Anglia, Norwich.

Summary

This submission focuses on the local water impacts of data centres in the context of mounting climate stress and growing pressure on water resources. It argues that data centres constitute a new, place-based water demand, with risks shaped by cooling demand, seasonality, drought sensitivity, siting, and cumulative clustering, especially in water-stressed regions. While some facilities use lower-water or alternative cooling approaches, current UK policy provides only a partial response because reporting is inconsistent and planning frameworks do not yet adequately address location-specific and cumulative risks. This submission recommends stronger reporting, better integration of data centres into water and spatial planning, greater use of lower-water cooling and non-potable supplies, and clearer siting criteria and constraints.

1          Introduction to author/organisation and reason for submitting evidence

1.1        Author/organisation

Dr Yi He, Associate Professor in Hydrology, Tyndall Centre for Climate Change Research, School of Environmental Sciences, University of East Anglia, Norwich.

The Tyndall Centre has significantly advanced the fundamental analysis of emission reduction from all major energy sectors, the understanding of climate impacts, risks, and adaptation options, the public perceptions of climate change, and the governance of climate negotiations and policymaking. The Tyndall Centre was founded in 2000 and is the oldest interdisciplinary research centre in the UK.

The University of East Anglia is internationally renowned and is in the UK top 20 for research quality (REF2021). The School of Environmental Sciences is one of the broadest and longest established academic departments in Europe focused on environmental research. It was ranked 1st in the UK for research quality in Geography and Environmental Sciences in The Times University Rankings 2025.

1.2        Reason for submitting

This submission focuses on the direct cooling-water impacts and the implications for UK water stress, drought resilience and siting. Dr He is an expert in hydrology, water resources, and climate change impacts and adaptation, with experience in assessing drought risk, hydrological extremes, and place-based water stress in the UK.

2          Key messages

Data centres constitute a new, place-based water demand under conditions of mounting climate stress. Their sustainability depends not only on how much water they use, but where they are located, when demand occurs, and the source of supply.

Impacts are concentrated rather than uniform. A small number of large or clustered sites can create localised pressure, particularly in already water-stressed regions.

Peak summer demand is the critical risk. Cooling demand is likely to coincide with periods of lowest water availability and highest system stress.

Current evidence is limited by inconsistent reporting. The UK lacks a reliable, standardised dataset on data-centre water withdrawals, consumptive use, source type, and peak demand.

Future risks are likely to grow in water-stressed regions. Southern and Eastern England are especially vulnerable due to existing deficits and increasing drought risk.

Existing policy frameworks are not yet sufficient for the scale and pace of data-centre growth. Water resources, drought and land-use planning do not yet provide a coherent, data-centre-specific approach to balancing seasonal demand and cumulative impacts in siting decisions.

Water availability should be treated as a constraint on data-centre expansion in some regions. Better alignment is needed between digital infrastructure planning and water availability at regional and catchment scale.

3          Evidence in response to the inquiry questions 2, 5 and 10

Figure 1. Conceptual representation of water-related impacts of data centres in the UK, supporting Q2 (impacts), Q5 (policy and regulations) and Q10 (competition and limits). Data centre locations are based on Data Centre Map (https://www.datacentermap.com/united-kingdom/, accessed 27 March 2026), which recorded 509 data centres in the UK, with the largest concentrations in London (202), Manchester (28), Cardiff (21) and Birmingham (20).

Figure 1 illustrates how cooling-related water demand (including withdrawals and consumptive use) interacts with water availability and environmental limits at the catchment and Water Resource Zone (WRZ) scale. Risks are shaped by both seasonal pressures (e.g. heatwaves and drought) and the geographic concentration of data centres, which can lead to localised and cumulative impacts on water resources and ecosystems. These dynamics underpin the key risks, policy gaps, and resource competition issues discussed in the following sections.

3.1        Q2. Environmental impacts (water)

The water-related environmental impacts of different types of data centre are driven mainly by their cooling approach: some designs rely on water-based/evaporative cooling and have higher consumptive water use, while others rely more on air-based cooling and have lower direct water demand. The water footprint of data centres is dominated by direct cooling-water use, with an additional indirect component linked to electricity generation. This submission focuses on the direct cooling-water impacts and the implications for UK water stress, drought resilience and siting.

3.1.1      Current water-related impacts in the UK

The main current water-related impact of data centres in the UK is cooling-water demand. For environmental assessment, the most important distinctions are between withdrawals and consumptive use (evaporative loss), and between potable public supply and non-potable or reclaimed sources (Mytton, 2021). These distinctions matter because environmental risk depends not simply on annual volumes, but on local water stress, the timing of demand, and the vulnerability of already stressed water systems (GDSA, 2025).

A robust national assessment remains difficult because measurement and disclosure are inconsistent. The GDSA (Government Digital Sustainability Alliance) report Water use in AI and Data Centres notes that there is currently no reliable dataset on the quantity of resources used by data centres, and only a minority of operators actively track water-use metrics, limiting benchmarking and environmental assessment across the sector (GDSA, 2025).

Emerging evidence from England suggests that impacts are concentrated rather than uniform. Dingle et al. (2026) estimate that English data centres may take around 1.879 million m³/year from public water supply, but that this demand is highly skewed: the top six sites account for 65% of sector consumption, while 67% of sites use less than 1,000 m³/year. This suggests that current pressure may be dominated by a relatively small number of very large and potentially more water-intensive facilities.

This broad pattern is consistent with the techUK (2025) survey of 73 English sites, which found that 51% of surveyed data centres used waterless cooling systems and 64% used less than 10,000 m³/year. However, that survey was voluntary and anonymous, focused mainly on commercial wholesale and colocation facilities, and did not capture the wider population of on-premise and enterprise data centres. It suggests that many commercial sites make limited use of potable water, but does not imply that data-centre water use is insignificant for the sector as a whole.

Even with limited national quantification, current risks are clearly spatially uneven. Where data centres are located in or near water-stressed areas, cooling demand can intensify local supply-demand pressure and increase the likelihood of tension with environmental flow objectives, particularly during heatwaves (Environment Agency, 2025; GDSA, 2025).

Current impacts are also seasonal. Periods of highest cooling demand are likely to coincide with periods when water systems are already under greatest stress from hot, dry weather. GDSA (2025) emphasise the sensitivity of UK water systems to these high-demand, low-supply episodes and the need for better forecasting and management of emerging demands from AI and data centres.

3.1.2      Future water-related impacts in the UK (short-, medium- and long-term)

Short term (to ~2030) – Peak-day operational stress:

In the short term, impacts are likely to be most visible through planning and operational stress during heatwaves and drought restrictions, particularly where cooling depends on potable water supply. The GDSA report emphasises that future AI and data-centre demand is “highly uncertain” and difficult to plan for without better data and reporting, increasing the likelihood of reactive rather than anticipatory management during dry years (GDSA, 2025; Dingle et al., 2026). Dingle et al. (2026) further note that water companies are more concerned about peak-day demand than annual totals, because peak cooling demand from data centres is likely to coincide with higher domestic demand and hotter weather. As a result, competition for water may become most acute during summer heatwaves and drought periods, when river flows, supply headroom, and ecological resilience are already under greatest stress.

Medium-term (to ~2040) – Cumulative regional and catchment pressure:

In the medium term, a key risk is cumulative impact where multiple facilities are developed or expanded within the same WRZ or catchment, reducing supply–demand headroom and increasing vulnerability during drought years. This risk should be understood in the wider context of rising regional water pressures already identified in national planning. The Environment Agency’s National Framework shows that the estimated additional public water-supply need between 2030 and 2055 is driven not by one factor alone, but by the combined effects of drought-resilience requirements, population growth, environmental improvement obligations, and climate change. These pressures are especially large in Water Resources South East and Water Resources East, where future deficits are among the highest in England (Environment Agency, 2025). In this context, data-centre expansion may not be problematic only because of the demand of individual sites, but because it can add a new, spatially concentrated and potentially peak-sensitive demand into regions already facing tightening water balances. This supports assessment at WRZ and catchment scale, alongside stronger integration between water-resources planning and spatial planning, rather than reliance on site-by-site appraisal alone (Environment Agency, 2025; GDSA, 2025).

Long-term (to ~2050 and beyond) – Peak demand coinciding with water scarcity:

Over the longer term, climate change is likely to increase the frequency and severity of hot, dry conditions, raising the likelihood that peak cooling demand coincides with periods of water scarcity. Evidence from a UK national-scale hydroclimate assessment (He et al., 2026) indicates intensification of drought-relevant indicators under warming: the national median maximum consecutive dry days increases from 32 in the baseline period to 36 at 2°C warming and 41 at 4°C warming, with particularly marked increases in southern and southeastern England. The same study suggests that streamflow drought conditions intensify most strongly in southern and eastern regions, including Anglian and Thames, with regional mean SSI1 (1-month Standardised Streamflow Index) approaching or falling below -0.5 at 4°C warming in some regions. In some catchments, values approach -1.0, indicating the potential for moderate to severe hydrological drought. These are also regions already characterised by high water demand and seasonal deficits, so future warming is likely to exacerbate existing pressures on water supply, river ecology, and drought resilience. Reyniers et al. (2023) similarly found that drought risk increases across almost all of Great Britain as temperatures rise globally, while southern and eastern England are particularly exposed to concurrent and prolonged drought under high-emissions scenarios. Taken together, this suggests that future data-centre water risks are likely to be greatest where sector growth overlaps with regions facing worsening low-flow and drought conditions.

3.1.3      Conclusion to Q2

Overall, the evidence indicates that the water-related impacts of data centres in the UK are likely to be geographically concentrated rather than uniform, with the greatest risks arising where large or clustered facilities are located in seriously water-stressed areas and where cooling demand peaks during hot, dry periods. UK-wide quantification remains limited because reporting is inconsistent, but emerging evidence from England suggests that a relatively small number of sites may account for a disproportionate share of sectoral water use and seasonal peak demand. Future impacts are therefore likely to become more significant where sector growth overlaps with increasingly drought-prone regions and where inadequate reporting continues to constrain effective assessment and management.

3.2        Q5. Are current regulations sufficient?

3.2.1      How should the impact of data centres be factored into future policies?

Data centres should be treated as a new form of place-based water demand under climate stress: their sustainability depends not only on how much water they use, but where, when, and from what source.

Current UK policy provides only a partial framework for managing these impacts. Existing instruments, including water resources planning, drought planning, abstraction controls, and land-use planning, can address some aspects of water use, but they do not yet provide a consistent, data-centre-specific framework for assessing cooling-water demand, seasonal peaks, cumulative impacts, or the implications of siting in water-stressed areas. Dingle et al. (2026) conclude that water-resources planning timelines are misaligned with the pace of data-centre growth, while GDSA (2025) argues that current plans to address future water deficits do not adequately account for emerging infrastructure demands such as data centres.

A central weakness is the lack of mandatory, standardised and location-specific reporting. The GDSA report, citing the Business & Human Rights Centre (2025), notes that there is currently “no reliable data on the quantity of resources used by data centres”, and that only two-fifths of operators actively track water-use metrics. This concern is also reflected in Privette et al. (2026), who argue that major gaps in transparency around data-centre water use undermine effective regulation, innovation and community planning. Dingle et al. (2026) likewise recommend centralised reporting of water use, source type and peak demand, broadly aligned with the EU Energy Efficiency Directive framework. The National Engineering Policy Centre (2025) reinforces this by calling for mandatory reporting of data-centre water consumption and withdrawal, water sources, and related environmental metrics. Without such information, local planning authorities, regulators and water companies cannot compare proposals consistently, assess cumulative and seasonal pressures, or plan effectively at regional and catchment scale.

A second weakness is the limited integration of water considerations into planning and infrastructure policy. GDSA (2025) highlights that the concentration of data centres in regions such as Greater London overlaps with areas already classified as seriously water stressed, creating predictable conflict hotspots. A third weakness is the unresolved tension between promoting data centres as Critical National Infrastructure and ensuring sustainable water use. Without stronger environmental conditions, there is a risk that policy support for AI and digital growth could weaken the ability of regulators and water companies to manage scarcity and protect public supply and environmental flows.

Data-centre water demand should therefore be factored into future policy through four linked changes: measurement, planning, screening, and conditional controls.
(1) mandatory reporting of withdrawals, consumptive use, source type, reuse fraction, and monthly and peak demand;
(2) explicit inclusion of data centres in water company water resources management plans and regional water-resources planning;
(3) a water-stress siting screen covering regional and local water stress, drought-plan restrictions, ecological low-flow sensitivity, and non-potable supply options; and
(4) drought-management requirements for sites with material cooling-water demand, especially where potable supply is proposed.

3.2.2      How important is the location of data centres, and what factors should be considered for optimum siting?

Location is fundamental to the sustainability of data centres because their water-related impacts are highly place-specific. The same facility may create limited risk in one area but significant pressure in another, depending on local water availability, drought sensitivity, ecological constraints, and the cumulative effect of other demands.

Optimal siting should therefore favour places with greater water headroom, lower drought sensitivity, lower ecological risk, and realistic access to non-potable or recycled supplies. Assessment should also consider whether proposed cooling demand is likely to coincide with peak summer stress, whether multiple facilities are clustering within the same planning or water-resource area, and whether the proposed cooling system would rely heavily on potable supply or involve substantial consumptive loss.

While some data-centre functions benefit from proximity to major interconnection hubs and concentrations of demand, this does not justify concentrating all future growth in already water-stressed regions such as Greater London. A more spatially differentiated approach is possible, with functions that require very fast response times located closer to major network hubs, while less time-sensitive or more water-intensive facilities are directed towards areas with suitable infrastructure and lower water stress.

In practice, optimum siting should balance at least six factors:
(1) regional and local water stress;
(2) drought resilience and likely restrictions during dry years;
(3) ecological sensitivity, especially low-flow and river-health constraints;
(4) the availability of non-potable, recycled or alternative water sources;
(5) cumulative pressure from clustering with other data centres or large users; and
(6) operational requirements such as power availability, fibre connectivity, resilience, and response-time (latency) needs.

Both GDSA (2025) and Dingle et al. (2026) indicate that sustainable data-centre growth depends on moving beyond site-by-site approval and considering water availability and environmental limits at wider planning and catchment scale.

3.3        Q10. Competition for water and implications for siting and sustainability

The resource demands of data centres, particularly for cooling, have the potential to intensify competition for water with other sectors, especially during hot, dry periods and in already stressed regions. The key issue is not only annual water demand, but where demand is located, when it occurs, and how reliable supply must be.

As discussed in Section 3.1.1, data-centre water demand is unlikely to be evenly distributed across the sector. Although many sites use relatively little potable water, a smaller number of large or more water-intensive facilities may account for a disproportionate share of demand and seasonal peaks. This means that competition for water is also likely to be localised rather than uniform.

Water companies are likely to be more concerned by peak-day and peak-season demand than by annual totals alone, because cooling demand can rise at the same time as domestic demand increases and weather conditions become hotter and drier. During droughts and heatwaves, when river flows, reservoir headroom and ecological resilience are already under pressure, this can create greater tensions with households, other businesses, and environmental flow requirements.

These tensions are likely to be strongest in regions where water stress is already significant. GDSA (2025) highlight that the concentration of data centres in regions such as Greater London overlaps with areas classified as seriously water stressed, creating predictable conflict hotspots. They further warn that siting water-intensive infrastructure in already water-stressed areas can transform a technical water-management issue into one of social equity and environmental justice.

There are also implications for decarbonisation. Cooling choices involve water–energy trade-offs, and some lower-water approaches may increase electricity demand, while electricity generation itself has an upstream water footprint. The implication is not that there is one universally optimal cooling solution, but that cooling design and water sourcing should be appropriate to local water stress, energy constraints, and future climate conditions.

From a sustainability perspective, this reinforces that water should be treated as a potential limiting factor for data-centre expansion in some regions. Future policy should therefore prioritise siting in locations with lower water stress, greater use of non-potable or recycled supplies, low-consumptive cooling approaches where feasible, cumulative assessment for clusters, and drought-mode operating plans for sites with material cooling-water demand. Without such measures, data-centre growth can exacerbate existing tensions between economic development, public water supply, ecological protection, and climate resilience.

4          Recommendations

R1     Minimum reporting standard: Require data centres with material cooling-water demand to report cooling-related water withdrawals and consumptive use, source type (potable/non-potable/reclaimed), and reuse fraction, using a consistent national template.

R2     Seasonal and peak demand: Require reporting of monthly totals and peak demand periods so drought-season pressures are visible rather than hidden within annual averages.

R3     Potable water justification: Where potable water is proposed for cooling, require an explicit alternatives assessment (e.g., reclaimed water, closed-loop designs) and a justification proportionate to local water stress.

R4     Water-stress siting screen: Embed a planning siting screen that considers Water Resource Zone (WRZ) stress, drought plan restrictions, and environmental flow sensitivity before consent is granted.

R5     Cumulative assessment for clusters: Where multiple data centres are proposed within the same WRZ or catchment, require a cumulative water impact assessment covering combined demand and drought-season effects.

R6     Drought-mode operating plan: Require a drought-mode operating plan for data centres with material cooling-water demand, setting out trigger conditions, demand-reduction actions, alternative supplies/reuse options, and monitoring arrangements.

5          Conclusion

Data centres are an emerging and place-based water demand whose impacts depend on location, timing, scale, and cooling technology. Current evidence suggests that impacts are concentrated rather than uniform, with the greatest pressures arising where large or clustered facilities overlap with already water-stressed regions and where cooling demand peaks during hot, dry periods. Better reporting, water-stress-informed siting, cumulative assessment, and drought resilience measures are therefore needed to ensure that future data-centre growth does not exacerbate existing tensions between economic development, public water supply, and ecological protection.

References

Business & Human Rights Centre (2025) Experts urge for mandatory reporting of AI data centers' energy & water use amid environmental concerns, Available online: https://www.business humanrights.org/it/ultime-notizie/experts-urge-for-mandatory-reporting-of-ai-data-centers-energy-water-use-amid-environmental-concerns/ (accessed 18 March 2026)

Dingle H, et al. (2026) WRc Final Report – Water Efficient Data Centres, Available online: https://www.wrcgroup.com/uploads/Water%20Efficient/Water%20Efficient%20Data%20Centres.pdf

Environment Agency (2025) The National Framework for Water Resources 2025: Water for growth, water for nature, water for a resilient future. Bristol: Environment Agency. Available online: https://assets.publishing.service.gov.uk/media/685d4a2ac2633bd820a92a99/2025_EA_National_Framework_Water_Resources_-_summary_document.pdf

GDSA (2025) Report – Water use in AI and Data Centres (Executive summary). Available online: https://assets.publishing.service.gov.uk/media/688cb407dc6688ed50878367/Water_use_in_data_centre_and_AI_report.pdf

He Y, et al. (2026) Escalating Hydroclimatic Extremes and Volatility in the UK Under 2 °C and 4 °C Warming. Earth’s Future, in review (minor corrections).

Mytton D (2021) Data centre water consumption. npj Clean Water. doi.org/10.1038/s41545-021-00101-w

National Engineering Policy Centre (2025) Engineering Responsible AI: foundations for environmentally sustainable AI. Available online: https://nepc.raeng.org.uk/media/2aggau2j/foundations-for-sustainable-ai-nepc-report.pdf

Privette AP, et al. (2026) Data Centers Water Footprint: The Need for More Transparency, AGU Advances. doi.org/10.1029/2025AV002140

Reyniers N, et al. (2023) Projected changes in droughts and extreme droughts in Great Britain strongly influenced by the choice of drought index. Hydrology and Earth System Sciences. doi.org/10.5194/hess-27-1151-2023

techUK (2025) Report – Understanding data centre water use in England. Available online: https://www.techuk.org/resource/techuk-report-understanding-data-centre-water-use-in-england.html

Page 10 | 10