Every time you tap a button to generate an image, get an instant summary, or ask for a detailed report from a language model, an unseen process is set in motion. In the world of artificial intelligence (AI), these seemingly effortless actions come at a cost—a significant environmental one. While AI is often championed as a tool for solving our most pressing climate challenges, its own operational footprint, particularly in terms of electricity and water consumption, is becoming an urgent issue in its own right.
This post will delve into the hidden environmental impact of AI, exploring the vast amounts of energy and water required to power its infrastructure. We’ll look at the issue from a global perspective, with a specific focus on the UK’s ambition to become a global AI leader and the sustainability challenges that come with it.
The Power-Hungry Brain: AI's Electrical Demands
AI models, especially the large language models (LLMs) like those that power generative AI, are incredibly computationally intensive. This is due to two main processes: training and inference. Training is the process of building the model by feeding it vast datasets, which can take weeks or months and consume immense amounts of energy. However, it's the ongoing process of inference—the energy used for every single query or command from a user—that is set to drive a continuous, exponential rise in electricity demand.
The scale of this demand is staggering. The International Energy Agency (IEA) has projected that global electricity demand from data centres could more than double by 2030, largely driven by AI. This would be equivalent to the entire current electricity consumption of Japan (IEA, 2025). To put this into perspective, a single ChatGPT query can use around 10 times more electricity than a standard Google search (Electric Insights, 2025). While this may seem trivial on an individual basis, when multiplied by billions of daily interactions, the total energy consumption becomes monumental.
For the UK, this presents a particular challenge. According to a House of Commons Library briefing, data centres currently consume around 2.5% of the UK’s electricity, a figure that is expected to rise four-fold by 2030, with AI being a key driver (House of Commons Library, 2025). This rapid increase in demand places immense pressure on an electricity grid that is already grappling with the transition to renewables and meeting existing energy security needs.
A Drop in the Digital Ocean: AI's Water Footprint
The environmental impact doesn't stop at electricity. All that computational power generates enormous amounts of heat, and data centres must be kept cool to prevent server malfunctions. This is where water comes in. Many data centres, particularly the older and larger ones, rely on evaporative cooling systems. This process is highly effective at managing heat but consumes vast quantities of water, as the water used to cool the equipment is evaporated into the atmosphere.
The numbers are sobering. A 2025 report published by Öko-Institut and other researchers projects that AI could be responsible for a global water withdrawal of between 4.2 to 6.6 billion cubic metres by 2027. To put that into perspective, that's more than half of the UK's total annual water withdrawal (Öko-Institut, 2025).
Global tech giants have already reported a significant increase in their water consumption, with companies like Google and Microsoft attributing the rise directly to their growing AI operations. A recent report from Google, referenced by WHEC.com, offered one of the most transparent estimates yet, stating that a single text prompt to its Gemini model uses approximately 0.26 millilitres of water for cooling—about five drops (WHEC.com, 2025). While this may seem small, the cumulative impact of billions of daily queries adds up to a staggering amount of water.
The UK is particularly vulnerable. The Environment Agency has forecast that England could face a daily water deficit of nearly 5 billion litres by 2050, due to a combination of climate change and population growth (GOV.UK, 2025). The additional strain from a burgeoning data centre sector, especially in areas already classified as "seriously water stressed", could exacerbate this crisis. For example, many of the UK's data centres are concentrated in the South East, a region already facing water scarcity issues.
The UK's Dilemma: Ambition vs. Sustainability
The UK government has made its ambition clear: to be a global leader in AI and digital technology. A new report from the University of York, in collaboration with Singapore Management University and Durham University, highlights a critical paradox: the very infrastructure powering our "digital green" initiatives carries a substantial carbon and environmental footprint (University of York, 2025). This research underscores the need for a comprehensive approach to building a digital economy that ensures long-term, sustainable success.
The government's new AI Opportunities Action Plan aims to fast-track the build-out of AI infrastructure through measures like new AI Growth Zones (AIGZs), which will offer streamlined planning approvals and priority access to clean energy (GOV.UK, 2025). This signals a strong commitment to expanding the sector. However, the House of Commons Library notes that high energy prices and grid capacity constraints are already limiting growth, creating a clear tension with the country's existing Net Zero goals (House of Commons Library, 2025). The creation of a dedicated AI Energy Council is a step towards managing this demand, but it remains to be seen how effectively it can balance ambition with the realities of a strained energy grid and finite water resources.
The Other Side of the Coin: AI as a Climate Solution
While AI's growing footprint is a serious concern, it is vital to acknowledge the other side of the paradox: AI's immense potential to accelerate the fight against climate change. A new study by the Grantham Research Institute on Climate Change and the Environment at the London School of Economics (LSE) finds that AI could reduce global emissions by 3.2 to 5.4 billion tonnes of carbon-dioxide-equivalent annually by 2035 (LSE, 2025).
The report highlights several key areas where AI can drive significant reductions, far outweighing its own energy consumption:
- Optimising Energy Systems: AI can manage and balance electricity grids, improving the efficiency of renewable energy sources like wind and solar by up to 20% (LSE, 2025).
- Encouraging Behavioural Change: AI-powered tools can provide tailored recommendations to help individuals and businesses make greener choices, from fuel-efficient driving routes to reducing food waste.
- Speeding Up Innovation: AI can be used to model and discover new materials for batteries and other clean technologies, a process that would otherwise take decades.
The research argues that AI's ability to transform complex systems like power, transport, and food is a crucial lever for the net-zero transition, but this will only be realised with strategic public policy and governance.
Towards a Greener AI: Solutions and Innovations
The good news is that the industry is not standing still. Many are recognising the need to address these issues head-on. There are several promising solutions and innovations on the horizon:
- Technological Efficiency: The development of more energy-efficient AI models and specialised hardware is a priority. Companies are also exploring advanced cooling methods, such as immersion cooling or liquid cooling, which use less water and are more energy-efficient than traditional air-cooling systems. A recent report by techUK, in collaboration with the Environment Agency, found that many UK data centres are already using minimal water, with 51% of surveyed sites using entirely waterless cooling systems (techUK, 2025). This shows that sustainable practices are possible and are already being adopted by some industry players.
- Renewable Energy Integration: Powering data centres with renewable energy sources is a critical step towards a greener AI. By ensuring that the energy consumed comes from solar, wind, or other low-carbon sources, the industry can significantly reduce its carbon footprint.
- Policy and Transparency: There is a growing call for governments to introduce policies that encourage sustainable practices. The University of York research points to the need for clear, supportive policies that provide guidance and incentives for a sustainable transition (University of York, 2025). This could include mandatory reporting of energy and water usage by data centre operators and regulations that disincentivise the use of potable water for cooling in water-stressed areas. The recent release of data from Google, while a welcome step, also highlights the need for a unified standard of measurement and public reporting across the industry to ensure full transparency and accountability (WHEC.com, 2025).
Conclusion: A Call to Action
The environmental impact of AI is real and growing, but it is not an insurmountable obstacle. The "invisible footprint" of our digital world, with its insatiable demand for electricity and water, is a powerful reminder that every technological advance has consequences. As a society, we must ensure that our ambition for AI leadership is matched by an unwavering commitment to sustainability.
Ultimately, the future of AI is not just about what it can do for us, but how we choose to build and power it. By holding the industry accountable, encouraging innovation, and demanding transparency, we can ensure that the rise of AI does not come at the expense of our planet’s resources.
As AI becomes an integral part of our lives, are we ready to hold the industry and ourselves accountable for its invisible footprint?
Sources
- Electric Insights. (2025). The Government's AI plans will supercharge electricity demand.
- GOV.UK. (2025). Report: Water use in AI and Data Centres Executive summary.
- GOV.UK. (2025). APPENDIX: AI and Data Centre Water Footprint in the UK: Strategic Recommendations for UK Government.
- House of Commons Library. (2025). Data centres: planning policy, sustainability, and resilience.
- International Energy Agency (IEA). (2025). Energy and AI.
- London School of Economics (LSE). (2025). New study finds AI could reduce global emissions annually by 3.2 to 5.4 billion tonnes of carbon-dioxide-equivalent by 2035.
- Öko-Institut. (2025). Environmental Impacts of Artificial Intelligence.
- techUK. (2025). Understanding data centre water use in England.
- University of York. (2025). The untold story of the environmental impact of AI and big data.
- WHEC.com. (2025). Is energy use for AI impacting the environment?.
- Yale E360. (2024). As Use of A.I. Soars, So Does the Energy and Water It Requires.
A video from YouTube offers an overview of the debate around AI's energy consumption. AI and the energy required to power it fuel new climate concerns
More Resources
Examining the potential benefits and risks of using artificial intelligence to advance global sustainability.
Drones with night vision are tracking elephant and rhino poachers in African wildlife parks and sanctuaries; smart submersibles are saving coral from carnivorous starfish on Australia's Great Barrier Reef; recycled cell phones alert Brazilian forest rangers to the sound of illegal logging. The tools of artificial intelligence are being increasingly deployed in the battle for global sustainability. And yet, warns Peter Dauvergne, we should be cautious in declaring AI the planet's savior. In AI in the Wild, Dauvergne avoids the AI industry-powered hype and offers a critical view, exploring both the potential benefits and risks of using artificial intelligence to advance global sustainability.
Dauvergne finds that corporations and states often use AI in ways that are antithetical to sustainability. The competition to profit from AI is entrenching technocratic management, revving up resource extraction, and turbocharging consumption, as consumers buy new smart devices (and discard their old, less-smart ones). Smart technology is helping farmers grow crops more efficiently, but also empowering the agrifood industry. Moreover, states are weaponizing AI to control citizens, suppress dissent, and aim cyberattacks at rival states.
Is there a way to harness the power of AI for environmental and social good? Dauvergne argues for precaution and humility as guiding principles in the deployment of AI.