When people think of "the cloud," they imagine something weightless and immaterial. In reality, the cloud is made of concrete, steel, copper, and silicon. It is housed in massive buildings, some exceeding one million square feet (the equivalent of more than 17 football fields), consuming as much electricity as entire cities and drawing millions of gallons of water every day to keep its servers from overheating. The environmental footprint of this infrastructure is substantial, growing rapidly, and largely invisible to the billions of people who use cloud services daily.
The IEA's Energy and AI report estimates that data centres consumed approximately 415 terawatt-hours (TWh) globally in 2024. This figure is projected to more than double to approximately 945 TWh by 2030, slightly more than Japan's total annual electricity consumption today. The United States alone hosts over 4,000 data centres, with a third concentrated in just three states: Virginia (643), Texas (395), and California (319). As artificial intelligence drives an unprecedented expansion of computing infrastructure, the environmental and community impacts of this growth demand serious attention from sustainability professionals, policymakers, and the public.
The Scale of Energy Consumption
Water: The Resource Nobody Talks About
Data centres require enormous quantities of water for their cooling systems. Server racks generate intense heat, and evaporative cooling is among the simplest and most cost-effective methods of temperature management. The critical distinction is that this water is not simply used and returned: much of it evaporates in the cooling process, meaning it is permanently removed from the local water cycle.
The World Economic Forum reports that the average data centre consumes 300,000 gallons of water per day. Larger facilities can consume up to 5 million gallons daily, equivalent to the water usage of a town of 10,000 to 50,000 people. A study by the Houston Advanced Research Center and the University of Houston found that data centres in Texas alone will use 49 billion gallons of water in 2025, potentially rising to 399 billion gallons by 2030. In Georgia, residents have reported problems obtaining drinking water from their wells after a data centre was constructed nearby.
Land Use and Community Impact
The physical footprint of data centres extends far beyond their building perimeters. Northern Virginia, described as "the data centre capital of the world," hosts approximately 300 facilities across a handful of counties, with two-thirds of the world's internet traffic passing through the region. Loudoun County alone has over 27 million square feet of data centre space, and officials expect data centre property tax revenues to approach $900 million in fiscal year 2025, nearly matching the county's entire operating budget.
However, the economic benefits come with significant trade-offs. Data centres consume farmland and open space. Thousands of on-site diesel backup generators, each the size of a rail car, create air quality concerns. The World Resources Institute reports that more than $64 billion in data centre projects were delayed or cancelled between May 2024 and March 2025 due to organised community opposition. One analysis in Virginia estimated that limited backup generator use could be associated with nearly $300 million in annual public health costs and 14,000 asthma-related health impacts across multiple states. In Indiana, Google withdrew a data centre proposal after residents opposed the rezoning of more than 450 acres, concerned about water and electricity consumption with few local employment benefits.
Carbon Emissions: A Growing Sector in a Decarbonising World
The IEA estimates that data centre emissions could reach 1% of global CO2 emissions by 2030 in its central scenario, or 1.4% in a faster-growth scenario. While these percentages may appear modest in isolation, data centres are one of the few sectors where emissions are projected to grow over this period, alongside road transport and aviation. Most other sectors are expected to decarbonise. Google, Microsoft, and Meta have all reported substantial emissions increases driven by data centre expansion in recent years, complicating their net-zero pledges.
The electricity consumption of the four largest data centre operators more than tripled between 2018 and 2023, from approximately 35 TWh to over 110 TWh. Training a single large AI model such as ChatGPT is estimated to generate 552 tonnes of CO2, equivalent to the annual footprint of 121 US households. While training is the most energy-intensive phase, AI inference (the process of running models after training) accounts for up to 90% of a model's lifecycle energy consumption.
What Can Be Done: Pathways to Sustainable Data Infrastructure
The environmental impact of data centres is not inevitable in its current form. Several states, including California, Illinois, Minnesota, New Jersey, and Virginia, have considered legislation requiring data centres to draw power from renewable sources and to disclose electricity and water usage. Cornell University research showed that smart siting, faster grid decarbonisation, and operational efficiency improvements could reduce data centres' carbon impact by approximately 73% and water impact by 86% compared to worst-case scenarios.
Advanced cooling technologies such as direct-to-chip liquid cooling and immersion cooling can dramatically reduce both water consumption and cooling energy demand. Battery energy storage systems can replace diesel backup generators, eliminating localised air pollution. And AI itself can be part of the solution: Google's DeepMind demonstrated a 40% reduction in data centre cooling energy through AI-powered optimisation. The critical requirement is transparency. A review of 31 Virginia municipalities with data centres found that 25 (80%) had non-disclosure agreements in place with operators, limiting public access to information about resource consumption and environmental impact.
Conclusion
The cloud is not weightless. Every email sent, every AI query answered, every video streamed depends on physical infrastructure that consumes electricity, water, and land at industrial scale. The growth trajectory of this infrastructure, driven by artificial intelligence, is one of the most significant energy developments of the decade. For sustainability professionals, understanding the environmental footprint of data centres is no longer optional. It is essential context for any credible assessment of corporate emissions, digital strategy, and the net environmental impact of AI adoption. The solutions exist: renewable energy procurement, advanced cooling, transparent reporting, and community-centred planning. The question is whether the pace of implementation can match the pace of expansion.
Share this post