Written submission from Water UK (DCU0037)

Data Centres’ Water Use - Water UK’s response to the Environmental Audit Committee call for evidence 02.04.26

Summary

Water UK is the trade association for the UK’s water industry. We represent all regulated water and wastewater companies in England, Northern Ireland, Scotland and Wales.

Government’s ambitions on data centres are at risk for no other reason than a lack of water. The UK Government aims to treble data centre capacity by 2030.[1]

Its Artificial Intelligence Strategy depends on being able to build data centres. Yet, despite the extraordinary growth envisaged and desired, the government has all but forgotten that most data centres require water to cool their systems. Mechanical air cooling, unlike evaporative cooling, does not need water, but it is much more expensive to operate and requires even more electricity (which is also in short supply).

The government’s forecasts for England’s future water needs also explicitly exclude data centres. This is a problem because such official forecasts cannot, in practice, be exceeded by water companies as they plan for investment in projects like new reservoirs. Water companies therefore have a limit placed on how much future water they can provide - a limit based on an inaccurate forecast that we already know to be much too low.

This problem is growing yet the government’s planning for artificial intelligence appears to assume the country will have sufficient supplies of water. The Science, Innovation and Technology Secretary’s policy for “Delivering AI Growth Zones” is dedicated to accelerating grid connections and bringing down energy prices.[2] It makes not a single mention of water. Neither does the government’s AI for Science Strategy.[3] Nor does its AI Opportunities Action Plan.[4] There appears to be an assumption that the country will always have enough water for its economic needs. Nothing could be further from the truth.

Right now, houses are being blocked and businesses are unable to expand for no other reason than a lack of water. The situation is so bad that, solely due to a lack of water, the Environment Agency objected to a cancer hospital being built in Cambridge. Water supplies are so low that it has forced a moratorium on business expansion in East Anglia and parts of Essex. A lack of water is the only thing stopping development in parts of Sussex, Suffolk and Lincolnshire. The problem is spreading.

Despite this worsening position, the government forces water companies to plan on the basis that businesses will need less and less water. Worse, companies are forced to build new water infrastructure with very little headroom between what is needed and running out. Headroom is only 2% in many parts of the country.

If government wants to achieve its data centre capacity ambitions, it must act now. Short of radical changes, including the urgent funding of new infrastructure and granting highly accelerated planning approval, it is almost certain that the country’s needs for data centres will be thwarted.

Immediate options available to government are:

• Reduce the limits on water abstraction: Government should review recent caps to water companies’ abstraction licences which severely limit the amount of water they can take from the environment. Water companies could have these caps suspended temporarily, to allow them to enable data centre and other growth, while they develop new water supply options. Many of these limits were imposed based on high level regional computer models rather than detailed environmental investigations. The Environment Agency has not reviewed any of these reductions despite evidence in water stressed parts of the country that some of them have not delivered the forecast environmental improvements.
• Allowing developers to fund new water supplies: Government should reform the developer charging regime to allow water companies to recover the full cost that data centres impose on the network, particularly as data centres can impose large network costs by adding significant pressure during their peak usage – likely to be hot summer days, which are also peak times for other users. Once data centre developers can see these costs, they can make an informed decision about investment in water efficiency measures onsite.
• Require water efficiency standards for data centres: The government should follow legislatures internationally by introducing minimum water efficiency standards for data centres, or for requiring, where possible, data centres to use non-potable water, either recycled from wastewater treatment works or through installing rainwater harvesting systems.

Water UK has not responded to every question in this consultation, but has indicated where it is responding to a question by including the question number in brackets. Throughout the response we have highlighted lessons from outside of the UK on the water usage of data centres (q14).

1. Context – increasing risk of water shortages

30 years of underinvestment

Not a single new reservoir has been built in England for more than 30 years despite a more than 20% increase in population over that time and significant changes to our climate.[5] Attempts by water companies to build new reservoirs have been consistently blocked by politicians and regulators in favour of keeping water bills down. Notable (and very expensive) examples include:

• In 2011, the then Environment Secretary, now Dame Caroline Spelman, rejected an application by Thames Water to build a new reservoir in Abingdon. It would have cost £1 billion had it been approved then, but she said there was “no immediate need”. In 2024, Thames Water was finally given approval, but costs over the wasted 13 years have now risen to between £5.5 billion and £7.5 billion.
• In 2014, Bristol Water applied to build a new reservoir at Cheddar for £114 million. It was rejected then, but was approved in 2024. The exact same reservoir will now cost £839 million. 

Historically, government and regulators required water companies to prepare for a ‘worst historic drought’, a drought with a 1 in 100 year probability of occurring. This low bar made making the case for new reservoirs difficult. However, in 2018 the National Infrastructure Commission highlighted that the cost of reactive measures in the event of a 1 in 500 year drought – such as tankering water around the country – would be twice as expensive (£40 billion compared to £21 billion) as proactive drought resilience through building reservoirs.[6]

Government has now accepted the need for greater drought resilience and in 2024 approved development funding for nine reservoirs, alongside construction costs for one at Havant Thicket, nine desalination plants, nine major transfers of water from wetter to drier regions and seven water recycling projects. Due to excessive planning and environmental restrictions, they will take years longer to build than they should. Given they are based on flawed forecasts, which significantly and deliberately underestimate the country’s need for water, all of these projects will not give business and people enough water.

Water is needed for life and increasingly business, but official forecasts are fatally flawed

Despite some infrastructure having been finally approved, we are facing growing instances of water shortages which will inhibit the growth of new businesses including data centres as a result of existing policies (q6).  They include:

1. Government targets which assume business water demand will fall: Water companies are being required to plan on the assumption that businesses will use less water over time, as the UK Government has set a target for businesses in England to reduce their use of water by 9% by 2038 and 15% by 2050. In attempting to justify these targets when they were set in 2022, the then UK Government asserted that, “...the reduction in economic dependence on heavy industry and manufacturing means there is no longer a strong link between non-household water demand and economic activity”.

Only three years later, the National Framework for Water Resources, the Environment Agency’s estimate of future water supply demand deficits - admits that “uncertain growth in some non-household sectors may mean demand increases beyond the levels planned.”[7] Indeed, it rightly acknowledges that “Data centres are fundamental to growth, industry and society,” and “Data centres can require a large amount of water for cooling processes and there is the potential for a new large demand as more data centres are built”. Yet, precisely because Water Resource Management Plans are required to factor in the flawed 2022 demand target, insufficient action has been authorised by regulators to provide new water, which means we will not have enough water for future data centres. This situation is almost certain to rapidly deteriorate. The situation is even worse because, as the Framework acknowledges “current non-household demand forecasts in [Water Resources Management Plans] are largely based on local authority plans which do not fully reflect the potential additional water needs of existing and new businesses”.[8]

Despite the UK Government’s classification of data centres as Critical National Infrastructure and its plans to establish AI Growth Zones with, reportedly, over 200 applications received from local authorities, the Water Resources Framework excludes gigafactories and data centres from its forecasts and only makes limited consideration of the water needed for the chemicals industry and navigation.

Yet, this Framework is the basis for determining whether new reservoirs should be built. It should not be acceptable that our National Framework for Water Resources dismisses these concerns by saying: “Instead, data centres will need to look to other sources of water for supply, for example use of water recycling or collaboration with other water users”.[9] Our economy cannot afford for businesses to be unable to grow because of such flawed estimates of their need for water. Hoping the businesses that need new data centres can “collaborate” with other users, which also need water, is no way to plan future water resources. This situation is resolvable in the future if we move away from these flawed assumptions and build sufficient headroom into the system to enable growth.[10]

Water companies are required by law to prioritise household over non-household customers, so this inadequate planning restriction will hit business growth first.

2. Water companies being required to plan on the basis of demand and leakage targets not forecasts: The Environment Act 2021 also set targets for reducing total water put into supply per head in England. This target assumes not only the reductions in business demand outlined above, but also large reductions in domestic consumption per head.  Water companies have very few levers to meet those targets and, on current trends, they will not be met, yet water companies are required to plan as if those targets will be met. As a result of planning water resources on the basis of targets that appear impossible to meet, demand is 220 million litres a day higher than was as assumed in Water Resources Management Plans.[11] Currently the deficit is being met by accepting a greater risk of environmental harm, and increased restrictions on water use in the event of a drought. This deficit may not be apparent in a wet year, where there is excess water to take from the environment. However, in 2025, which saw the driest spring in England since the nineteenth century, this deficit ultimately meant that reservoir, river and groundwater levels dropped earlier, restrictions on use occurred earlier, and there was greater environmental harm than would otherwise be the case.
3. A misjudgement about the balance of risk: The Environment Agency’s approach to ‘headroom’ – allowing for more water to manage uncertainty around future forecasts for factors such as climate change and population growth – in water resources planning guidelines is flawed. It assumes companies should accept more risk into the future, rather than allowing for more headroom in their plans, but it does not acknowledge that building infrastructure to increase the supply of water takes between 10 and 15 years. As such, it could well be too late to provide the ‘headroom’ we need if forecasts are wrong – which they almost certainly are. The impact of having too little water for the environment, society and the economy, is far higher than the cost of providing too much.
4. A lack of information on data centre water use: Every five years, the Environment Agency publishes its National Framework for Water Resources which makes assumptions concerning the needs businesses will have for water in the future. The most recent National Framework was published in June 2025 did not include an estimate of the water demands of data centres, and instead suggested that ‘data centres should look to other supplies of water beyond using public water supply, for example using recycled water’. While Water UK agree that data centres should both use non-potable water and be as water efficient as possible, as set out below, this is not the status quo and a lack of water due to ineffective planning should not be a barrier to infrastructure which is critical to the government’s growth strategy.

2. Impact of data centres

2.1             Data centres’ water use

Current estimates suggest that data centres’ water use globally is likely to more than double by 2050 from 2025 levels.[12] This growth will not be evenly distributed, with services-based economies like the UK likely to see a greater proportion. The UK has a data centre development pipeline which is almost double that of the next highest in Europe – Spain.

Data centres can use water for domestic-like uses – drinking water and toilet facilities for any onsite staff – but the largest volumes of water are likely to be required for cooling computer chips used for processing information. When the external temperature is too hot to be cooled simply by ventilating the building, water can be used for cooling. There are two categories of cooling[13]:

• Open loop cooling: Water is passed through a data centre to cool chips and heated water then evaporates as it passes through cooling towers. This method tends to be the least water efficient with consumptive water use of 1.85 litres per kilowatt hour of data processing.
• Closed loop cooling: Water is either used to wet a media through which air passes before it cools chips (adiabatic cooling) or liquid cools chips directly with heat subsequently ejected through a secondary cooling system. These systems tend to be much more water efficient, with water usage being as low as 0.1 litres per kilowatt hour of data processing.

Several factors influence how a data centre is designed and the volume of water it uses.

Cost

As the cost of water is significantly below the cost of energy, all else being equal, data centre operators will try to use less energy and more water. This incentive is likely to be particularly strong in England where decades of underinvestment have kept water bills artificially low, but where energy prices are among the highest in the developed world.  Medium sized businesses in the UK pay around 92% more than the EU median for electricity.[14]

Temperature

The amount of water data centres use is highly dependent on temperature.  In North America in July, data centres use around 2500% the volume of water they use in January.[15] This is because below certain temperatures data centres can rely on ambient outside air temperatures for cooling, but they cannot do this in the summer months, when ambient temperatures are too warm. These figures are not directly translatable to the UK because average summer temperatures in the UK are seven degrees cooler than those in the US.

2.2             Data centre water use in the UK

Data centres currently use approximately 6.6 million litres of water per day. (q2 and q3)[16] It is equivalent to the water used by 20,000 average homes.[17] With the government’s ambition to treble data centre capacity by 2030, this could increase to 19.8 million litres, equivalent to 60,000 homes.[18]

While this is only 0.14% of the 14 billion litres put into supply every day, peak demand – on hot summer days, when demand across the rest of the network is high – can be more than 8 times higher than average annual demand. Water UK estimates that average peak demands imposed by current data centre capacity is 14 million litres per day, or 7 million litres per gigawatt of capacity.[19] If the government achieves its ambition to treble existing capacity, it could increase to 42 million litres of water per day – twice the daily capacity of the reservoir currently under construction at Havant Thicket in Portsmouth.

The above figures only represent the average water use per gigawatt of power consumed. If all new data centres performed in line with the least water efficient data centres – which operates at around 15.75 million litres of water per gigawatt of capacity, then the peak strain on the network would increase to 77 million litres of water per day – almost four times the daily capacity of Havant Thicket – and equivalent to the average daily usage of more than 235,000 homes.[20] While this is a worst case scenario, and data centres may be expected to increase their water efficiency per gigawatt of processing capacity over time, water companies have received requests for access to up to 21 million litres of water per day from individual data centre developers in the last two years. This difference in current observed peak demand – using data from summer 2025 – and future requested peak demand is likely to be due to the sector planning for higher average and peak summer temperatures during the asset lives of the data centres – measured in decades - as a result of climate change.

[21][22]The above features of data centre water use present several policy and regulatory challenges for the management of water systems.

3. Policy challenges created by data centres

The nature of data centres’ water use creates two related policy challenges:

• Data centres are concentrated in areas that are already water stressed: More than three quarters of data centres are located in the already water stressed east and south east of England.[23] This is because many data centres need to  be located near financial centres to ensure ‘low latency’ – meaning that financial transactions are as close to instant as possible. While data centres may not impose large additional pressures on the network, in terms of annual average usage, the water networks in this region have very limited headroom – and in some places, such as Cambridge and North Sussex, houses and business expansion has been blocked or delayed. Any new data centre growth in these areas may be impossible without policy change because of existing pressure on the network. As leakage and demand targets are already very ambitious, the only solution to deliver new supply is to increase the amount of water taken from the environment – which may be harmful – or to build new supplies, and major reservoirs can take 15 years to plan and build. This means that in the short term either government prevents data centres from being built in these areas or data centres eat into headroom, preventing other businesses from opening or housing from being built.
• Peak demand pressure: Data centres pose specific challenges because, as set out above, they may request very large volumes of water which they only use on very hot summer days. Unfortunately, those hot days are also peak demand times for all other uses of water. Water companies manage daily and hourly fluctuations in demand – we use less water on cooler days, and in the middle of the night – by storing 2-3 days average supply of treated water in service reservoirs to manage peaks. If peaks increase, as they will need to accommodate data centres, then service reservoir and/or treatment works’ treatment capacity will also need to increase. Failing to do this would mean risking supplies for all customers during hot summer peaks.

4. Policy Changes

There are a set of shorter- and longer-term policy changes to ensure there is sufficient water to address the above challenges to enable the data centre revolution.

4.1             Short term – implementable in the next year

In the short term data centres should be as water efficient as possible and only be located in areas of water stress where this is essential to their function or where to do otherwise would be prohibitively expensive. This is because if data centres are situated in areas of water stress when they do not need to be, or use more water than they need to, this will deplete headroom and limit growth of businesses that do need to be location specific. To enable these discussions, more clarity will be needed on actual intended water use at the planning stage.

Short term measures to enable data centres without hampering economic growth in other sectors of the economy include:

• Disclosure of Water Usage at planning stage: This is an effective tool for enabling data centres by standardising information requirements on water use. The European Union’s Energy Efficiency Directive, passed in September 2025, requires data centres operating in the EU to disclose their Water Usage Effectiveness – litres of water used per kilowatt hour of data processing.[24] In the United States, data centre regulation is occurring at state level rather than federally, but several states, including California, Iowa and Michigan are drafting legislation requiring data centres to submit information on water usage. Disclosure using this consistent metric up front will enable water companies to understand what investment they might need to make in their network to accommodate this demand.
• Requirement to use non-potable water where possible: Where possible, data centres should be using non-potable water. If a data centre is located near a wastewater treatment works, this could be tertiary treated final effluent from the works, but it could also be provided through a rainwater harvesting system on site. This could also be supported through greywater recycling where possible. In Singapore, since the start of January 2024, businesses which use more than 60 megalitres of water a year are required to recycle at least 50% of the water they use.[25]
• Water efficiency standards for data centres in water stressed areas: In water stressed areas, data centres should be as water efficient as possible to avoid limiting growth in other areas of the economy. This could be encouraged through, for example, a minimum water usage efficiency standard. This would not need to apply to all data centres, as there is a potential trade-off between water usage and operational cost, but could apply specifically in areas already classified as ‘water stressed’ for the purposes of enabling compulsory metered charging.[26] In some jurisdictions, governments are drafting more radical legislation, with South Carolina and Kansas both considering legislation banning open loop cooling for data centres, and Singapore aims for all data centres to achieve a Water Use Efficiency of 2.0 or less within the next 10 years.[27]
• Review of restrictions on the amount of water that can be abstracted: In some parts of the country, a shortage of water is driven as much by recent caps to abstraction licences as it is by climate change or a growing population. In some of these cases, these caps could be temporarily suspended to enable economic growth – such as supplying data centres – while new infrastructure is developed. Many of these cuts to abstraction are not based on environmental investigations of the sites in question, and instead are based on high level computer models. There is also currently no mechanism for water companies to challenge these cuts if they discover that they do not deliver the environmental benefit originally envisaged.

4.2             Medium term – implementable this parliament

In the medium term, i.e. this parliament, government could support data centres through funding companies to expand their networks. Ofwat sets a five year price control defining companies’ investment over that period. At the end of 2024, Ofwat confirmed £104 billion in investment between 2025-30. However, Ofwat’s assessment was based on company business plans drawn up in 2023 and Ofwat have recognised that both population growth, and the government’s ambition for development driven economic growth has increased since then. As a result, Ofwat has proposed a ‘re-opener’, a window for companies to apply for additional investment, to support growth. Ofwat should fund new short and longer-term water resource options through its price control growth re-opener. At present, the re-opener contains unnecessary hurdles, including that the needs case for new schemes cannot acknowledge that current planned for growth is insufficient as it is based on optimistic assumptions about demand reduction.

4.3             Long term – implementable for the next price review

In the longer term the water resources planning process should be reformed:

• The Environment Agency should change water resources planning guidelines now so water companies can plan on the basis of realistic forecasts and not ambitious targets for demand and leakage reduction.
• The Environment Agency should require water resources management plans have pathways to adapt if growth is higher than expected within an investment period to avoid companies having to wait five years to adjust forecasts. The annual review of the plans could then function as a decision point for switching pathways.
• The Environment Agency should require plans to have a greater level of headroom so that water resource zones can enable unplanned or speculative development, without a lack of water becoming a barrier to growth.

10