On 16 October 2025, BBC News revealed that data centres powering artificial intelligence (AI) in Scotland are consuming enough tap water annually to fill 27 million half-litre bottles. The data, obtained through Freedom of Information requests, shows that the volume of tap water used by Scotland’s data centres has quadrupled since 2021.
Read the original article on BBC News.
As AI technologies like OpenAI’s ChatGPT and Google’s Gemini surge in global popularity, so too has the resource consumption of the data centres powering them. BBC analysis found that Scotland’s 16 existing data centres—and many more in planning—use significant amounts of electricity and water to cool servers that handle AI workloads.
Scottish Water described this growth in water use as “significant,” though it currently represents only about 0.005% of Scotland’s total supply. Experts from the University of Glasgow warned that water use could soon become environmentally unsustainable if the number of data centres continues to expand without new cooling technologies or alternative water sources.
Why this is important to take note of
This story highlights a critical, often overlooked environmental cost of AI. While discussions around artificial intelligence typically focus on ethics, data privacy, and automation, the physical footprint of AI infrastructure—especially its use of energy and water—poses new sustainability challenges.
For a global audience, this is a wake-up call: as AI adoption accelerates worldwide, the demand for data processing power—and therefore cooling resources—will rise exponentially. Without responsible environmental management, AI progress could conflict with international climate and net-zero commitments.
Similar stories from around the world
- United States: Reports earlier this year found that Google’s data centres in Iowa used nearly 1.5 billion gallons of water in 2023, sparking local concerns about drought and sustainability.
- Netherlands: The Dutch government temporarily halted construction of a Meta (Facebook) data centre in Zeewolde due to energy and water consumption fears.
- Ireland: The Irish planning authority imposed stricter environmental regulations on new data centres after warnings that they could consume 30% of the national electricity grid by 2030.
The real problems and issues
The fundamental issue lies in resource inefficiency—the fact that current “open-loop” cooling systems used in most Scottish data centres rely on a continuous supply of potable tap water. This not only wastes drinking water but also increases the carbon footprint due to energy needed for pumping and treatment.
A lack of transparent reporting from data centre operators further compounds the issue. Companies rarely publish data on their environmental impact, making it difficult for regulators and researchers to assess sustainability performance.
What went wrong?
Regulators, developers, and operators alike underestimated the environmental externalities of AI infrastructure. Policymakers failed to establish clear reporting mandates or sustainability standards for data centres, while consumers and AI developers overlooked the resource intensity behind the tools they use daily.
Many major technology firms and private investors running or funding Scottish data centres failed to disclose their water and energy usage data, limiting public accountability. Most continue to rely on inefficient “open-loop” cooling methods rather than sustainable systems.
Even where companies are exploring “closed-loop” alternatives, these are not yet widely implemented, and environmental monitoring remains voluntary.
They lacked:
- Robust environmental impact assessments (EIAs) for water and energy use.
- Mandatory sustainability reporting frameworks.
- Closed-loop or hybrid cooling systems that recirculate water.
- Policies linking data growth to net-zero targets.
However, they did have:
- High-efficiency servers and renewable energy commitments in some cases, but insufficient integration of water-management strategies.
How these issues could have been prevented or fixed
- Early adoption of closed-loop cooling systems to recycle and minimise water usage.
- Locating data centres near wastewater treatment plants, allowing use of treated effluent instead of potable water.
- Mandating environmental transparency, with regular public disclosure of water and energy consumption data.
- Incentivising green data infrastructure through government grants or tax credits tied to sustainability performance.
- Embedding AI sustainability standards into national digital strategies—ensuring growth does not undermine environmental goals.
Why this should matter to you, not just policymakers
At its heart, this story isn’t just about data centres or AI—it’s about how invisible technologies shape our visible world. Every chatbot conversation, photo filter, or AI-generated image is tethered to a physical process consuming water, electricity, and raw materials somewhere on the planet.
When we talk to an AI, we rarely think about the human consequences of our convenience—the rivers diverted to cool servers, the power grids strained to train models, or the local communities who may face water scarcity as a result.
Technology doesn’t exist in a vacuum; it exists in our shared environment. The question is not whether we should build more data centres, but whether we can do so responsibly, equitably, and transparently. We need innovation that respects the boundaries of the natural world and recognises that sustainability is not a luxury—it’s a duty.
If AI is to truly serve humanity, then it must also serve the planet that sustains us.
What can you do?
We understand it might be difficult to do, but the only thing that can be done at this time is to try to stop using AI powered searching when it really isn’t needed.
Other than that, information leaders, companies and organisations can also think about their data footprint. Mainly, getting rid of their Redundant Obsolete and Trivial (ROT) matter.
Recently we hosted a webinar Your Data Carbon Footprint: What It Is & Why It Matters and invited Prof. Thomas Jackson, Information & Knowledge Management, and Prof. Ian Hodgkinson, Strategy Research, both from at Loughborough University Business School to speak to our audience of records management professionals and information leaders.
Your Data Carbon Footprint: What it is and why it matters
This webinar explores the often-overlooked environmental impact of digital data, delving into the concept of the data carbon footprint and its implications for sustainability.
As data storage and processing account for 2.5%-3.7% of global greenhouse gas emissions – surpassing even the aviation industry – understanding and mitigating digital carbon is crucial.
This discussion also examines the role of AI and its data-driven methodologies in amplifying these environmental challenges.
Tom introduces key problem areas, such as “dark data” and storage inefficiencies, and discusses actionable strategies, including the Data Carbon Scorecard, to help you and your organisations manage your data’s environmental footprint.
Our Principal Consultant, Rachel Mitchell then:
- Discusses how this eye-opening information impacts records management professionals.
- Explains how you can use ‘data sustainability’ as a message to enhance your voice when it comes to promoting the importance of data and information lifecycle management to your companies and organisations.
- Explores the records management tools available and look at the skills framework needed to deploy them.
By fostering a responsible data and AI approach, we can support a global shift toward digital decarbonisation, aligning with net-zero goals.
This webinar is aimed at:
- Information governance junior and senior managers
- Records management junior and senior managers
- Sustainability champions from all industry sectors
- Project managers tackling records and information lifecycle issues.
