By Kalina Abdulla
Abstract
The rapid expansion of cloud computing and artificial intelligence has generated unprecedented global demand for data centers, which rely heavily on large-scale cooling systems to keep servers from overheating. This paper examines the environmental consequences of data center water consumption, with a focus on local ecosystems, and reviews existing responsible next-generation solutions rather than proposing new ones. Through an analysis of current literature, this study reveals that hyperscale data centers can consume millions of gallons of water daily, comparable to small municipalities, with particularly severe impacts in water-stressed regions. The research identifies impacts of freshwater extraction, including reduced streamflows, altered thermal regimes, and declining viability of aquatic ecosystems. While industry responses show promise through zero-water cooling technologies and closed-loop systems, inconsistent reporting standards obscure the full scope of environmental impacts. This paper advocates for mandatory water-use disclosure policies and strategic geographic planning to balance technological advancement with long-term ecological sustainability.
Introduction
Each time an AI chatbot is presented with a question, a video is uploaded to social media, or a file is saved to the cloud, somewhere on the planet the temperature rises slightly within a large warehouse filled with computers. Many envision the “cloud” as something intangible. However, the cloud is physical, housed in enormous windowless buildings packed with thousands of computers running nonstop and generating so much heat that they need constant cooling to prevent overheating. These facilities are cooled using water, requiring millions of gallons to operate.
These data centers are draining rivers, emptying underground aquifers, and competing with local communities for water supplies. In drought-stricken areas of Nevada and Arizona, where water is scarce, technology companies are pumping out freshwater at unexpectedly high rates. Meanwhile, fish populations are declining, wetlands are drying up, and local communities are watching their water sources disappear, largely to support streaming services and AI training. As digital activities increase and artificial intelligence advances, this hidden environmental crisis will continue to escalate. This paper explores the environmental impact of technology on water resources, the damage to ecosystems, and whether effective solutions can be implemented in time.
Literature Review
Research shows that modern data centers use large amounts of water because they rely on evaporative cooling and water-based chillers to keep servers from overheating (Nature, 2021). A single hyperscale facility can use hundreds of thousands to millions of gallons of water per day, which is similar to the needs of a small town (GRESB, n.d.). This demand is rising quickly as AI systems and cloud computing expand, creating even higher energy and water use (Environmental Law Institute, 2025). These withdrawals can harm local ecosystems by lowering streamflows, warming water temperatures, and reducing the amount of freshwater available for wetlands and nearby communities (Ceres, 2025). The problem is especially serious when data centers are built in drought-prone places like Nevada or Arizona, where water supplies are already stressed, meaning water sources are limited, droughts are frequent, and existing demand from cities, agriculture, and ecosystems already exceeds sustainable supply (The Guardian, 2025).
Although many data center operators and technology companies still do not publicly report their water use, making it difficult to understand the full environmental impact, this gap has led experts to call for mandatory disclosure policies (The Guardian, 2025). Encouragingly, there are already proven solutions in practice. Several data centers have significantly reduced water consumption through the use of recycled water, non-evaporative cooling systems, and local water-replenishment initiatives (Veolia, n.d.). In addition, leading technology companies are developing low-water or zero-water cooling systems, such as closed-loop cooling, chip-level liquid cooling, and seawater or geothermal cooling (Microsoft, 2024). Closed-loop systems reuse the same water in sealed pipes, greatly reducing evaporation losses, while chip-level liquid cooling improves heat efficiency by delivering coolant directly to server chips, lowering overall water demand. Seawater and geothermal cooling further reduce freshwater use by relying on ocean water or stable underground temperatures instead of limited local water supplies. Overall, while progress is being made, the literature shows that better planning, transparent reporting, improved technology, and government-enforced environmental regulations are essential to limit water overuse and prevent long-term ecological damage.
Methodology
Results and Discussion
The Mechanics and Scale of Data Center Water Consumption
Understanding why data centers use so much water starts with understanding how computers work. Modern servers generate massive amounts of heat when processing data because billions of tiny electrical signals move through transistors on computer chips as calculations are performed. As electricity flows through these components, some energy is lost as heat due to electrical resistance, and this heat builds up rapidly as servers run continuously at high speeds. Without cooling, servers would quickly overheat and break down. To solve this problem, many data centers use evaporative cooling systems or water-based chillers that transfer heat away from the servers using water as a cooling medium (Nature, 2021). In evaporative systems, water absorbs heat and evaporates into the air, taking the thermal energy with it, but this means the water is constantly used up and requires replacement.
Figure 1: Water consumption in a data center
The scale of water use in these facilities is enormous. A single large data center can consume hundreds of thousands to millions of gallons of water every day, which is about the same amount a small town would use (GRESB, n.d.). To put this in perspective, a facility using one million gallons daily consumes enough water to serve approximately 10,000 people based on typical household usage. As observed in Figure 2, the total amount of water used in 2025 reached 279 gigalitres (GL). The rise of artificial intelligence has amplified this problem. As companies invest in constructing larger facilities to keep up with demand for services, these sites account for a large share of local water consumption.
Figure 2: Projected growth in global data center energy use, water consumption, and resource demand driven by artificial intelligence, showing steep increases in electricity and water needs through 2030.
Data centers consume vast amounts of energy. AI hyperscale facilities often consume tens to over a hundred megawatts (MW) of power continuously, comparable to the electricity needs of 15,000–75,000 homes. At a global scale, data centers used approximately 410 terawatt-hours (TWh) of electricity in 2025, according to the International Energy Agency (IEA). The IEA also projects that global data center electricity demand will continue rising rapidly through 2030, largely driven by the growth of artificial intelligence and cloud computing. To better interpret this annual energy use, electricity consumption can be converted into average power demand using the formula:
Power (GW) = Energy (TWh) / Time (hours)
Using this formula to convert 410 TWh for one year (365 days = 8,760 hours) of use, the power consumption is 46.8 gigawatts (GW). As higher electricity use generates more heat, cooling systems must work harder, increasing water consumption. Therefore, as AI becomes more common in everyday applications, data center water use will likely continue to rise unless more efficient technologies are widely adopted.
At the national scale, overall data center demand is rapidly growing. Industry forecasts estimate that total U.S. data center electricity demand could exceed 100 gigawatts (GW) by 2035, a dramatic increase driven largely by AI workloads. To put this in perspective, one gigawatt can power roughly 750,000 homes, meaning future data center demand could rival that of major metropolitan areas. Some projections suggest that grid demand from data centers may nearly triple by 2030 if current trends continue (451 Research via S&P Global).
Figure 3: U.S. data center power demand and projection to 2035
In addition to their substantial energy consumption, these facilities cover a vast amount of physical space. A single hyperscale data center can occupy hundreds of thousands to millions of square feet. To put this in perspective, some of the largest data center complexes are nearly as large as a small city and, in extreme cases, approach half the size of Manhattan (The Guardian, 2025).
Figure 4: Comparison of the physical size of large hyperscale data center campuses with Manhattan, illustrating the massive land footprint of modern digital infrastructure.
These sprawling campuses can house tens of thousands of servers, operate 24 hours a day, and require complex infrastructure including backup generators, water treatment systems, and security perimeters. According to the Environmental Law Institute (2025), the growth of AI and cloud computing has caused substantial increases in both energy and water consumption. As these facilities scale up in both size and power demand, the strain on local electricity grids and water supplies becomes even more severe. The combination of enormous heat generation, continuous operation, and massive size means that cooling these facilities consumes millions of gallons of water daily, comparable to the water usage of a small town or city. As AI becomes increasingly integrated into daily life, data center water consumption will likely continue to rise unless changes are made.
Ecological Consequences and Geographic Vulnerabilities
The environmental harm caused by excessive water consumption by data centers occurs through interconnected hydrological and ecological impacts. Large withdrawals of freshwater from rivers, lakes, and groundwater sources reduce water availability for natural ecosystems and downstream communities.
Research by Ceres (2025) indicates that these withdrawals can lower streamflows, alter water temperatures in rivers and streams, and reduce water availability for wetlands. These changes place significant stress on aquatic ecosystems and threaten species that depend on stable flow conditions and temperature ranges for survival and reproduction, including fish, amphibians, and aquatic insects.
Lower streamflows concentrate pollutants, reduce dissolved oxygen levels, and eliminate shallow-water habitats critical for feeding and reproduction. Decreased water volume also accelerates warming, often exceeding the temperature limits of sensitive species. Collectively, these stressors can reshape ecosystems and reduce biodiversity.
Broader sustainability frameworks emphasize that environmental harm produces long-term consequences that extend beyond short-term economic gains. The Aga Khan Development Network’s framework emphasizes responsible stewardship of natural resources and long-term water security, offering a conceptual model relevant to sustainable data center water management.
The severity of these environmental impacts is strongly influenced by the geographic location of data centers. In water-stressed regions such as Nevada and Arizona, facilities draw from already limited water supplies, creating direct competition with agriculture, urban populations, and ecosystem conservation (The Guardian, 2025). The concentration of data centers in major cities worldwide illustrates that water and ecosystem pressures are a global concern.
Figure 5: Top cities hosting the world’s most data centers in 2025, illustrating global concentration in key metropolitan hubs.
Industry Innovation and the Path Toward Sustainability
Awareness of these environmental challenges has led the data center industry to develop more sustainable solutions. Major technology companies are investing in cooling systems that dramatically reduce or eliminate water consumption. Microsoft (2024), for example, has introduced data centers featuring zero water evaporation cooling designs that recycle water in a closed loop, conserving up to 125 million liters annually per facility.
Google and Amazon Web Services have committed to becoming water positive by 2030, using reclaimed or non-potable water in many facilities and investing in water replenishment initiatives (Data Centre Magazine, 2023). Closed-loop cooling systems, chip-level liquid cooling, and high-temperature water cooling designs further reduce both energy and water use (NVIDIA, 2024; Veolia, n.d.).
In addition to cooling systems, companies are implementing broader conservation initiatives, including recycled water use, drought-tolerant landscaping, and water risk assessments across supply chains. Despite these advances, reliance on legacy systems and inconsistent disclosure continue to limit widespread adoption. Environmental organizations have therefore called for mandatory water use reporting to improve transparency and oversight (The Guardian, 2025).
Counterarguments and Limitations
Although environmental concerns associated with data center water use are significant, some critics argue that mandating advanced cooling technologies could impose financial burdens on smaller companies. Certain solutions, such as seawater cooling, are geographically limited. In addition, incomplete data disclosure complicates efforts to fully assess industry-wide impacts. At the same time, data centers provide essential services such as remote work, online education, and telemedicine. The challenge is therefore not elimination, but sustainable operation that balances economic feasibility with environmental protection.
Discussion
The evidence shows that data center water consumption is a growing environmental challenge driven by AI and cloud computing expansion. Facilities that have implemented low-water and zero-water cooling demonstrate that viable solutions already exist. The primary barriers to adoption are economic, regulatory, and transparency-related rather than technological.
Governments can address these barriers by requiring standardized public water use reporting and incorporating water availability into data center siting decisions. Market pressure from consumers and businesses can further incentivize sustainability. As climate change intensifies water scarcity, the long-term viability of digital infrastructure will depend on reducing its environmental footprint.
Conclusion
Data center water consumption presents a serious challenge at the intersection of technological advancement and ecological responsibility. This paper shows that data centers consume water at scales comparable to entire towns, with measurable impacts on local ecosystems, particularly in water-stressed regions.
Advances in cooling technology demonstrate that near-zero water consumption is achievable, indicating that environmental harm is not inevitable but largely a result of outdated practices. Meaningful progress will require coordinated action from policymakers, industry operators, researchers, and consumers.
The digital revolution has transformed society, but ensuring that this transformation proceeds sustainably is one of the defining challenges of our time. Data center water consumption exemplifies this challenge while also representing a problem that can be effectively addressed through informed policy, investment, and technological innovation.
Reference List
Aga Khan Agency for Habitat. (n.d.). Water and sanitation. Retrieved January 30, 2026, https://the.akdn/en/how-we-work/our-agencies/aga-khan-agency-habitat/water-and-sanitation
Aga Khan Development Network. (n.d.). Environment and climate. Retrieved November 20, 2025, https://the.akdn/en/how-we-work/our-agencies/aga-khan-health-services/environment-and-climate
Aga Khan Development Network. (n.d.-b). Environment and climate change. Retrieved January 30, 2026, https://the.akdn/en/what-we-do/building-resilient-communities/environment-climate-change
Aga Khan Development Network. (n.d.-c). Our approach to development. Retrieved January 30, 2026, https://the.akdn/en/how-we-work/our-approach/our-approach-to-development
Aga Khan Foundation Canada. (2017). Policy for environmental sustainability 2015–2024. https://cooperation.ca/wp-content/uploads/2020/09/AKFC-Policy-for-Environmental-Sustainability-2015-2024_UPDATED-JULY-2017-1.pdf
Amplyfi. (n.d.). AI’s growing ecological footprint: Energy, water, and resource challenges. Retrieved January 30, 2026, https://amplyfi.com/blog/ais-growing-ecological-footprint-energy-water-and-resource-challenges/
Ceres. (2025). Drained by data: The cumulative impact of data centers on regional water stress. Retrieved November 30, 2025, https://www.ceres.org/resources/reports/drained-by-data-the-cumulative-impact-of-data-centers-on-regional-water-stress
Data Center Frontier. (2023, March 1). Google developing new climate-conscious cooling tech to save water. https://www.datacenterfrontier.com/cooling/article/33001080/google-developing-new-climate-conscious-cooling-tech-to-save-water
Data Center Frontier. (2024, December 26). Microsoft's ‘sustainable by design’ initiative seeks to transform data center water efficiency. https://www.datacenterfrontier.com/sustainability/article/55249649/microsofts-sustainable-by-design-initiative-seeks-to-transform-data-center-water-efficiency
Data Centre Magazine. (2023, November 28). Reclaimed wastewater to be used at 20 AWS locations. https://datacentremagazine.com/articles/reclaimed-wastewater-to-be-used-at-20-aws-locations
Environmental Law Institute. (2025). AI's cooling problem: How data centers are transforming water use. Retrieved December 10, 2025, https://www.eli.org/vibrant-environment-blog/ais-cooling-problem-how-data-centers-are-transforming-water-use
Fierce Network. (n.d.). NVIDIA has no chill: AI drives explosive data center cooling demand. Retrieved January 30, 2026, https://www.fierce-network.com/cloud/nvidia-has-no-chill
GRESB. (n.d.). Cooling data centers: Managing water use in the age of AI and ESG. Retrieved December 10, 2025, https://www.gresb.com/nl-en/cooling-data-centers-managing-water-use-in-the-age-of-ai-and-esg/
International Energy Agency. (2025). Global data centre electricity consumption, by equipment, Base Case, 2020–2030 — chart. https://www.iea.org/data-and-statistics/charts/global-data-centre-electricity-consumption-by-equipment-base-case-2020-2030
Microsoft. (2024, December 9). Sustainable by design: Next-generation datacenters consume zero water for cooling. Microsoft Cloud Blog. https://www.microsoft.com/en-us/microsoft-cloud/blog/2024/12/09/sustainable-by-design-next-generation-datacenters-consume-zero-water-for-cooling/
Nature Water. (2021). Data centers and water use (Article 101). https://www.nature.com/articles/s41545-021-00101-w
NVIDIA Corporation. (2024). FY2024 corporate sustainability report. https://images.nvidia.com/aem-dam/Solutions/documents/FY2024-NVIDIA-Corporate-Sustainability-Report.pdf
OilPrice.com. (2025). U.S. data center demand could hit 106 GW by 2035. Retrieved January 15, 2026, https://oilprice.com/Energy/Energy-General/US-Data-Center-Demand-Could-Hit-106-GW-by-2035.html
Pew Research Center. (2025, October 24). What we know about energy use at U.S. data centers amid the AI boom. https://www.pewresearch.org/short-reads/2025/10/24/what-we-know-about-energy-use-at-us-data-centers-amid-the-ai-boom/
S&P Global. (2025, October 14). Data center grid-power demand to rise 22% in 2025, nearly triple by 2030. https://www.spglobal.com/energy/en/news-research/latest-news/electric-power/101425-data-center-grid-power-demand-to-rise-22-in-2025-nearly-triple-by-2030
The Guardian. (2025, July 16). Zuckerberg says Meta will build a data center the size of Manhattan in latest AI push. https://www.theguardian.com/technology/2025/jul/16/zuckerberg-meta-data-center-ai-manhattan
The Guardian. (2025, December 4). Nevada AI data centers. https://www.theguardian.com/technology/2025/dec/04/nevada-ai-data-centers
Veolia. (n.d.). Data center cuts annual water consumption by more than 12 million gallons [Case study]. Retrieved January 5, 2026, https://www.watertechnologies.com/case-study/data-center-cuts-annual-water-consumption-more-12-million-gallons
