Executive Summary
- Founder and CEO of climate tech Showerkap, Steve Harding, explores how AI’s water problem isn’t about data centres, it’s more about what it’s being used for – the debate becomes nuanced if the water use is aiding in delivering environmental benefits that outweigh the facility’s resource footprint.
- A growing number of AI-enabled water management platforms are deployed, from digital twins to predictive analytics that can optimise schedules and reduce water loss and energy consumption.
- The question we should be asking is are the activities powered by those facilities justify the water consumption. If AI is helping reduce leakage, improve water efficiency, strengthen infrastructure resilience and accelerate the transition to a more sustainable economy, then the industry is generating environmental value that must be included in the equation.
Amid news of AI’s escalating water consumption, the real question is not how much water data centres use, but what it is being used for, says Steve Harding, founder and CEO of climate tech Showerkap
The debate around artificial intelligence and water consumption is rapidly gaining momentum. Recent warnings from the United Nations suggest that AI’s growing water demand could soon rival global drinking water needs. And in the US, reports highlight the expansion of data centres in drought-prone regions, raising concerns about the environmental cost of the digital future.
Water-stressed regions are becoming more common, heatwaves are intensifying and power grids are struggling to keep pace with soaring demand from AI workloads. With global water demand expected to exceed freshwater supply by 40% within the next four years, the resource footprint of data centres demands scrutiny.
Discussions around AI’s environmental impact often focuses exclusively on inputs – water, energy and carbon – while ignoring the outputs. Yet the value of any resource-intensive technology can only be properly assessed by examining both sides of the equation.
If an AI platform consumes water and energy to generate marketing content or summarise emails, then questions about resource efficiency are entirely justified. However, if that same infrastructure is being used to deliver environmental benefits that outweigh their resource footprint, eliminating water waste, optimising energy consumption or improving critical infrastructure resilience, the picture becomes far more nuanced.
Net impact matters
Consider the growing number of AI-enabled water management platforms now being deployed globally. Smart leak detection systems can identify hidden losses across water networks before they become catastrophic failures. Predictive analytics can optimise pumping schedules, reducing both water loss and energy consumption. Digital twins are helping utilities model future demand scenarios and improve network resilience.
In hospitality, behavioural technology is already delivering measurable results. UK pilot programmes have demonstrated that simple digital interventions and behavioural nudges can reduce shower water consumption by more than 50%, cutting both water use and the energy required to heat it.
In these cases, the cloud infrastructure and data centre capacity supporting the technology are not environmental liabilities. They are enablers of resource conservation because AI applications may consume water in order to save significantly larger quantities elsewhere.
The challenge is that these beneficial workloads are competing for infrastructure capacity alongside countless applications delivering far less societal or environmental value.
Data boom tools could help deliver solution
The growth trajectory of AI is extraordinary. Across the Gulf Cooperation Council region, data centre capacity is expected to triple by 2030 – according to business advisory firm FTI Consulting. All this comes at an environmental cost with the centres straining already scarce water resources and consuming 15 billion litres in Saudi Arabia alone last year.
In an area where temperatures regularly exceed 45 degrees centigrade, the job of cooling thousands of servers requires vast amounts of water – often drawn from energy-intensive desalination plants.
Data centre operators are already exploring alternatives, including direct-to-chip liquid cooling, closed-loop cooling systems and more efficient heat management technologies. Many are making genuine progress in reducing water intensity and improving overall efficiency. Rather than being forced to choose between AI and sustainability, experts claim that the tools driving the data boom could help solve the problem. Yet technological innovation alone will not solve the issue.
The stressed South
Regions like the UK’s South East will be officially classified as water-stressed by 2030, while leakage rates and consumption patterns remain stubbornly high. At the same time, regulatory expectations are tightening, and public tolerance for waste is rapidly declining.
As a nation of rain, many don’t associate the UK with water scarcity, yet the UK is ramping up for trials of timed showers and seasonal tariffs this summer to tackle shortages. Hotter, drier summers, growing population pressures and increasing regional imbalances in supply are converging into a new reality – drought risk is no longer hypothetical – it is structural and the need to protect this finite resource is more pressing than ever.
The policy response has focused on long-term resilience. New reservoirs, inter-regional transfers, desalination and leakage reduction are all coming into play. But these are slow solutions, often taking years – if not decades – to deliver and desalination can have negative impacts on marine ecosystems.
Meanwhile, thousands of smaller, distributed interventions – leakage reduction, smart metering, water reuse, efficiency in buildings and industry – remain under-funded despite delivering faster returns. These measures reduce demand at source, cut energy use and build resilience across the system and are equally worthy of both attention and investment.
Water scarcity is a systems problem
One of the more uncomfortable realities surrounding AI is that governance has struggled to keep pace with growth.
Much of today’s AI demand comes from widespread consumer and enterprise adoption. AI assistants are now embedded into productivity software, search engines, collaboration platforms and countless mobile applications. Users are increasingly encouraged to deploy AI for routine tasks, often without visibility into the associated resource consumption.
Collectively, these millions of seemingly insignificant interactions account for a substantial proportion of AI-related energy demand.
As policymakers and regulators begin examining the environmental footprint of AI infrastructure, the focus should move beyond absolute consumption metrics and towards value creation. Not all workloads generate the same societal benefit and not all resource use produces the same outcome.
The industry needs more transparency and a clearer understanding of which AI applications are delivering measurable environmental and social returns.
The question we should be asking
As scrutiny of AI’s water footprint intensifies, the debate needs to evolve. The question is not whether data centres consume water. They do. The more helpful question is whether the activities being powered by those facilities justify that consumption. If AI is helping reduce leakage, improve water efficiency, strengthen infrastructure resilience and accelerate the transition to a more sustainable economy, then the industry is generating environmental value that must be included in the equation.
Water scarcity is emerging as one of the defining constraints of twenty-first century growth and data centres sit at the heart of that challenge.
But rather than framing AI and sustainability as opposing forces, the industry should focus on ensuring that its growing computational power is directed towards solving the very environmental problems it is increasingly accused of creating. Artificial intelligence is uniquely positioned to identify inefficiencies, optimise resource allocation and reveal hidden opportunities for conservation at scale.
