海角视频

Water has long been the quiet partner in conversations about data centre expansion. Energy and land tend to dominate the headline debate, while fibre continues to shape decisions about location and viability. Yet water sits at the heart of how digital infrastructure functions.

Water use in data centres varies widely, but the volume involved is far from trivial, used primarily for cooling the systems. International evidence shows that large hyperscale facilities can consume up to 20,000m鲁 per day, comparable to the needs of a town of 10,000 to 50,000 people[1]. Even in the UK, where many sites already adopt more efficient or waterless technologies, water companies have experienced a series of requests of 25 l/s or more for data centre connections, equivalent to more than 750,000m鲁 a year. This places cooling squarely within the wider debate about long鈥憈erm water security. Potable water is perceived as the most reliable source for data centre operators, yet it is also the same supply on which households, agriculture and many industries depend. As demand for digital infrastructure grows, any increase in potable withdrawals will compete directly with other economic uses, especially in catchments already facing climate鈥憆elated stress.

What happens to the water once it is used is equally important. Evaporative and adiabatic systems can consume water outright, while other systems discharge substantial volumes of warmed or chemically treated wastewater that must be managed within existing sewerage and treatment capacity. The UK鈥檚 own assessments highlight that national water resource plans finalised in 2025 do not yet adequately account for the rapid rise in water demand associated with AI鈥慸riven data centre growth, creating a risk that unplanned cumulative withdrawals could intensify regional water stress.[2] In regions where wastewater treatment plants already operate near capacity, new high鈥憊olume cooling discharges may require network reinforcement or parallel investment to maintain compliance and protect river health.

Cooling is the primary driver of water demand. As Jack Foster, an associate in 海角视频鈥檚 water team, explains, the first major decision any developer faces, is the selection of cooling technology itself. 鈥淭here are a range of cooling technologies available, some with high demands for water, and then there are some that use much less water, all the way up to air-cooled systems. Generally, the ones that use much less water tend to be more expensive and more complicated, and the cheaper, simpler ones use more water.鈥�    

This fundamental trade鈥憃ff frames much of the strategic conversation around sustainable data centre growth. High鈥憌ater鈥憉se systems remain more cost鈥慹ffective and easy to implement, but their cumulative impact becomes significant when multiple developments cluster within a region or when long鈥憈erm demand rises, as predicted.

Governance challenge

These pressures sit within a broader infrastructure and governance challenge. Water is expensive to move, meaning supply and treatment capacity must align with the physical geography of proposed sites. While other planning domains such as drainage and flood risk are supported by well鈥慹stablished hierarchies, there is currently no equivalent regulatory framework guiding sustainable water use for data centres, leaving developers free to default to high鈥慶onsumption options if utilities are willing to supply them. Water鈥慹fficient cooling, hybrid systems and non鈥憄otable supply can significantly reduce demand, provided investment is planned early and integrated with regional strategies. The UK Government has begun to recognise the scale of the challenge, with calls from industry bodies such as the Government Digital Sustainability Alliance (GDSA) advocating for regulatory change,  but a coherent national policy for data centre water use has yet to be fully developed.

Every data centre relies on cooling, and every cooling strategy brings with it a series of choices that affect not only operational efficiency but environmental resilience, cost, public perception and long鈥憈erm capacity. As the UK prepares for accelerated growth in data鈥慽ntensive industries, understanding these choices has become essential for planners, local authorities, utilities, developers and operators.

A second foundational choice is the type of water used for cooling. Developers must decide whether to rely on potable supply, which is straightforward to access but competes with domestic demand, or to integrate non鈥憄otable sources such as recycled or treated effluent, which can significantly reduce pressure on local networks but require early planning and close coordination with utilities.

Fluid decision-making

Potable water is by far the simplest source to access: it is well distributed, well monitored and can be tapped directly from existing networks. As Jack notes, 鈥減otable water is very much the easiest because we鈥檝e got a good network which data centre developers can tap in to and get water supply just like any house or business does.鈥� Yet as Jack also points out, simplicity and sustainability do not always align. The volumes involved in large鈥憇cale cooling could, under rapid growth scenarios, place pressure on local water supply systems that were never designed to accommodate industrial鈥憇cale withdrawals in areas already experiencing stress from climate change, population growth and ageing infrastructure. For a local authority, it means weighing the convenience of potable supply against the risk that sustained high鈥憊olume withdrawals could erode the resilience of systems already under strain, with real consequences for public confidence and long鈥憈erm capacity.

Non鈥憄otable sources present an alternative, ranging from rainwater harvesting and recycled grey water to treated effluent from sewage works. These options dramatically reduce demand on potable systems but come with their own spatial and engineering constraints. Locating a data centre near a wastewater treatment plant, for example, can unlock stable access to an appropriate non鈥憄otable water source, yet such locations may not align with land availability, fibre convergence, energy capacity or local development priorities. Even when such locations are technically feasible, they rarely sit within traditional growth zones.

Again, from the local authority perspective, this means recognising that while non鈥憄otable sources can ease pressure on potable supplies, the spatial compromises required may limit viable sites and force difficult decisions about how data centre growth aligns with wider land鈥憉se and infrastructure priorities.

For local authorities, these choices translate into a set of difficult but unavoidable trade鈥憃ffs that shape long鈥憈erm resilience. A cooling strategy that relies heavily on potable water may be technically straightforward for developers, yet it can place additional pressure on systems already facing competing domestic and industrial demands, particularly in regions where water stress is rising and resource plans have historically underestimated future non鈥慼ousehold demand. Current national water resource plans finalised in 2025 do not fully account for the projected surge in water use linked to AI鈥慸riven data centre growth, creating a risk that uncoordinated withdrawals could intensify local scarcity and undermine public confidence in major developments.[3] At the same time, while some UK data centres now use efficient or waterless cooling, water companies have experienced a series of requests of 25 l/s or more for data centre connections, equivalent to more than 750,000m鲁 a year, a scale of use that can only be supported where infrastructure, treatment capacity and catchment conditions align. Local authorities are therefore being asked to balance economic opportunity with finite resources, ensuring investment does not compromise the long鈥憈erm security, affordability and social licence of the places they serve.

A further consideration is the regulatory position of water companies. Unlike domestic supply, there is no legal obligation for water companies to provide water for non鈥慸omestic uses, which means requests from data centre developers can be declined or negotiated on a case鈥慴y鈥慶ase basis. This introduces an additional layer of uncertainty for operators seeking to secure long鈥憈erm cooling capacity. However, the designation of data centres as Critical National Infrastructure gives government the ability to apply pressure where strategically important schemes face supply constraints. In practice, this can shift the balance in favour of approvals for key developments, but it also highlights the need for clearer frameworks that support transparent, sustainable and equitable decision鈥憁aking for all parties involved.

Spatial strategy

Transporting water over meaningful distances is rarely a viable solution. Unlike electricity, which can be transmitted efficiently across regions, water quickly becomes expensive to move. For planners, this has profound implications. Water is not just a question of supply; it is a question of spatial strategy, land use and infrastructure geography. Regions seeking to attract data centre investment must consider from the outset whether their locations can support sustainable water use without placing undue strain on existing systems. Our advisory teams are experienced at delivering feasibility studies to answer these kinds of big strategic questions. 

The UK鈥檚 planning landscape currently offers limited guidance. Other areas of planning, such as drainage or flood risk, are supported by well鈥慹stablished frameworks that guide developers through a hierarchy of options, requiring them to demonstrate why more sustainable or lower鈥慽mpact solutions are not feasible before moving to less favourable ones. A similar regulatory structure for data centre water use does not currently exist, leaving developers largely free to select high鈥慶onsumption, potable鈥憌ater鈥慸ependent systems if utilities agree to supply them. Jack Foster says he believes a comparable framework approach could be a simple and impactful solution.

For local authorities, adopting this kind of systems鈥憈hinking approach is essential, because only by aligning land, water, energy and infrastructure evidence can they steer developers toward solutions that balance growth with long鈥憈erm resilience and public trust.

The implications are not only technical but political. Public perception can shift rapidly when severe weather highlights the fragility of water systems. Hosepipe bans sit uneasily alongside the notion of industrial cooling systems drawing high volumes of potable water. Local authorities, therefore, face a dual responsibility: ensuring the long鈥憈erm resilience of their region鈥檚 water supply while safeguarding public trust in major development decisions. Clear communication, evidence鈥慴ased strategies and early engagement with water companies become essential.

Leading by example

In places where water considerations have been embedded early and treated as a core design constraint, data centres have demonstrated that growth and resilience can be compatible. Apple鈥檚 Viborg facility in Denmark is often cited as a leading example, applying the AWS Water Stewardship Standard to integrate catchment鈥憀evel understanding[4], efficient cooling design and responsible resource management into its operation[5]. Elsewhere, Los Alamos National Laboratory鈥檚 use of treated effluent for cooling, supported by targeted investment in filtration and pump upgrades, has significantly reduced potable鈥憌ater demand and increased reuse cycles[6], showing how engineered solutions can relieve pressure on local supplies. Facilities such as Facebook鈥檚 Prineville data centre[7] also demonstrate what is possible when location and technology are aligned, achieving significant reduction in water use through innovative outside鈥慳ir cooling systems.[8]

By contrast, regions where water risks have been underestimated or left unregulated offer a clear warning of the consequences. In the United States, several data centres have drawn heavily from community supplies, including one Iowa facility that consumed a billion gallons in a single year[9], while in some localities cooling demands have accounted for more than a quarter of community water use, often with limited transparency for residents or planners.[10]

Growing focus

Forward-thinking authorities in the UK are beginning to ask how many data centres their region can realistically accommodate and what resource constraints will shape that capacity.

鈥淯ltimately, water is a finite resource, and at some point, without proper constraints, there wouldn鈥檛 be enough to go around,鈥� Jack warns. For high鈥慻rowth regions, this is not a distant scenario. It is a foreseeable problem that needs addressing before cumulative failures dictate reactive policy.

Navigating this complexity means putting structured guardrails around decision鈥憁aking, with clear hierarchies that prioritise low鈥憌ater cooling and non鈥憄otable sources wherever feasible. It also requires early, place鈥憇pecific modelling that helps authorities weigh cumulative impacts, test alternatives and steer developers toward solutions that protect long鈥憈erm resilience while still enabling growth.

Ultimately, water is a finite resource, and at some point, without proper constraints, there wouldn鈥檛 be enough to go around

Jack Foster, an associate in 海角视频鈥檚 water team

海角视频鈥檚 water resource management teams are already deeply engaged in helping authorities and developers navigate these emerging constraints. Drawing on decades of modelling, masterplanning and strategic water allocation, we are adapting proven methodologies from residential and mixed鈥憉se sectors to the specific pressures of data centre growth. Frameworks already used for residential and mixed鈥憉se development can be adapted to data centre contexts. The questions are similar: What is the projected demand? What sources are available? How can non鈥憄otable supply be maximised? How can cumulative pressure be managed across whole water catchment areas? 

Water is no longer a background consideration. It is a strategic constraint and, if planned well, a potential driver of innovation in location strategy, technology choice and long鈥憈erm resilience. As data centre growth continues, regions that confront these questions early, grounded in evidence and supported by clear frameworks, will be those best placed to attract sustainable investment and maintain public confidence in the benefits of digital development.

This article is part three of a series. Read part four here: How fibre infrastructure is shaping the geography of data centre growth

Get in touch with Yalena Coleman, who is leading on data centre advisory for 海角视频, to continue the conversation.

---

[1] https://www.eesi.org/articles/view/data-centers-and-water-consumption

[2] :

[3] https://www.eesi.org/articles/view/data-centers-and-water-consumption

[4] https://a4ws.org/wp-content/uploads/2025/01/AWS_Water-Stewardship-in-Data-Centres_2025.pdf

[5] https://www.aquatechtrade.com/water-stories/industrial-water/apples-and-amazons-big-data-players-tackle-water

[6] https://www.osti.gov/biblio/2433993

[7] https://www.datacenterknowledge.com/cooling/facebook-revises-its-data-center-cooling-system

[8] https://www.nature.com/articles/s41545-021-00101-w.pdf

[9] https://theconversation.com/data-centers-consume-massive-amounts-of-water-companies-rarely-tell-the-public-exactly-how-much-262901

[10] https://www.ceres.org/resources/reports/drained-by-data-the-cumulative-impact-of-data-centers-on-regional-water-stress

---