The relentless expansion of artificial intelligence is placing an unprecedented and largely hidden strain on the UK’s essential resources, with new data revealing the startling scale of its energy and water appetite. Behind every chatbot query and generated image lies a physical infrastructure of power-hungry datacentres, whose environmental footprint is now triggering urgent regulatory and community responses.
A Looming Energy Crunch
Datacentres are the engine rooms of the AI boom, and their power demands are colossal. According to industry analysis, these facilities consume between 10 and 50 times more electricity per square metre than a typical commercial building. In 2023 alone, UK datacentres used an estimated 5.0 terawatt-hours (TWh) of electricity, accounting for 2% of total national demand. The trajectory, however, is what is causing profound concern. The National Grid predicts datacentre electricity demand could increase sixfold by 2034, potentially consuming 30% of the UK’s total electricity. Other forecasts suggest demand could more than quintuple to 26.2 TWh by 2030, representing nearly a third of all commercial sector power use.
This surge is being directly fuelled by AI. The International Energy Agency (IEA) notes that globally, datacentre power demand is growing four times faster than all other sectors, on track to exceed Japan’s entire electricity use by 2030. The computational intensity of AI is a key driver; experts state that generative AI models consume “orders of magnitude” more energy than traditional computing. Asking an AI chatbot a question uses far more power than a simple web search, a disparity likened by Oslo-based climate analyst Ketan Joshi to “driving to the shops in an SUV instead of riding your bike.”
The practical effect is a severe strain on the grid. A queue of approximately 140 proposed datacentre schemes, driven by AI demand, could collectively require 50 gigawatts of electricity—5 GW more than the country’s current peak demand. This has led to warnings that AI infrastructure could displace other critical projects, such as new housing, in the connection queue. In response, the UK government is consulting on reforms to prioritise “strategically important projects,” including AI datacentres and new “AI Growth Zones” intended to streamline planning where low-carbon energy is available.
The Overlooked Thirst of AI
While energy use dominates headlines, the vast water consumption of AI and its datacentres is a growing parallel concern, particularly for cooling systems. Globally, research published in the journal Patterns estimates AI’s water use could reach between 312.5 and 764.6 billion litres annually by 2025—a footprint similar to global bottled water consumption. Projections suggest AI could increase global datacentre water usage to 6.6 billion cubic metres by 2027, equivalent to over half of the UK’s total water usage.
Within the UK, a recent study found current potable water use by datacentres is approximately 1.88 million cubic metres per year, with just six facilities responsible for 65% of that total. A techUK report indicates many sites use efficient systems, with 51% using waterless cooling. However, the largest facilities can still consume millions of litres daily, and the indirect “embedded” water used in generating their electricity is a significantly larger, often overlooked, factor. Analysis of one proposed hyperscale campus suggested its indirect water use could be over 50 times its direct consumption.
With the UK facing a projected daily water deficit of 5 billion litres by 2055, this demand is triggering calls for stricter oversight. There is growing public pressure for transparency, with 88% of the UK public believing operators should disclose environmental impact reports. Reflecting this, the UK government is now being advised to introduce mandatory water-use reporting for datacentres.
Community Tensions and Economic Trade-offs
The physical manifestation of this digital growth is sparking local friction. While the datacentre sector is a major economic contributor—projected to deliver £4.7 billion in gross value added and support 43,500 jobs in the UK in 2024—its local impacts are increasingly contested. Communities are mobilising against proposed facilities, concerned about their drain on resources, land use, and the constant noise and light from 24/7 operations.
Dr Bronwyn Cumbo, a social researcher at the University of Technology Sydney, notes that these facilities often cluster in industrial hubs, and the “incentive to be a good neighbour really depends on the company.” There are calls for the industry to be held accountable. In Australia, a coalition including the Australian Conservation Foundation has proposed “public interest principles” demanding datacentres invest in new renewable energy and use water responsibly. Its chief executive, Adam Bandt, argues “Big tech corporations should be forced to do their fair share so they don’t drain our resources.”
This tension underscores a broader challenge: balancing AI’s economic promise with its environmental and social cost. Prof Jeannie Paterson, co-direct of the Centre for AI and Digital Ethics at the University of Melbourne, acknowledges that “We’re becoming immersed in this technology. It’s really hard to avoid,” but stresses that “how it’s part of our lives is something we can definitely control.”
The Path to a Sustainable Digital Future?
In the face of this soaring demand, the industry and regulators are exploring mitigation strategies. The shift to renewable energy is critical; renewables are the fastest-growing electricity source for datacentres, and through Power Purchase Agreements (PPAs), operators are becoming major buyers of green energy. Technological innovations like advanced liquid cooling and ambitious projects to reuse datacentre waste heat for local communities are also being pursued.
Furthermore, trials are examining how datacentres can act as “flexible loads,” temporarily reducing power use during periods of grid stress to aid stability. However, significant hurdles remain. A major barrier identified in the UK is a lack of transparency and measurement, with 56% of IT decision-makers reporting they struggle to accurately measure the emissions generated by AI workloads, hindering sustainability claims.
For individuals, the sense of being locked into an energy-intensive system is palpable. Ketan Joshi criticises tech giants for baking generative AI “deep into their systems,” a tactic he compares to “the growth of single-use plastics in the 1970s.” While complete opt-out is difficult, users can limit use by unsubscribing from platforms, tailoring web searches to exclude AI results, and avoiding it for unnecessary, energy-heavy tasks. Joshi views such acts not just as energy saving, but as a “meaningful act of resistance” and part of collective action against a “corrosive, harmful industry.”
The conversation, as Dr Cumbo observes, is coming to a head. The UK, and the world, must now decide how to power its AI ambitions without overheating the planet and draining its water supplies in the process.
