Snapping a photo with your smartphone. Downloading a podcast. The Google search that led you to this blogpost.
What do these all have in common? They all end up in the Cloud, our digital library floating in perpetuity. The Cloud allows us to save vast amounts of data, but it also requires large physical buildings that contain computer systems and the associated components. The role of these data centers in storing, managing, and distributing data has remained largely obscured from public view. Maintaining this insatiable information chamber uses vast amounts of energy and water, produces excess heat, and releases extra carbon into the atmosphere. The cloud may be invisible, its impact on the environment is not. As we stream more shows, take more pictures, send more emails, and use more chatbots, we must also find more approaches to technological sustainability.
As technology expands, so does the need for data centers. The expansion will increase to handle the estimated 29.3 billion devices by 2030 [1] and the 175 zettabytes of data created and stored by 2025 [2]. To put this increase into perspective, imagine stacking thick textbooks from Earth to Pluto and back multiple times. Each book represents a piece of information—a video, a photo, a document, etc. The sheer volume of data would fill countless libraries, stretching beyond the boundaries of our planet into outer space. The U.S. contains one-fourth of all global data centers [2].
How does data in the cloud affect water supplies?
Data centers need to be kept cool to house the thousands of computers that hold all our data. Typically, water is the preferred method to cool these massive facilities. Total U.S. data center water usage is 1.7 billion liters per day, compared to the overall U.S. daily water consumption of 1.218 trillion liters per day [1]. According to a Virginia Tech study, data centers used approximately 513 million cubic meters of water in 2018. Much of that water use comes from electricity use, but about a quarter from using water for direct cooling [10].
Virginia, particularly Northern Virginia, hosts the largest data center market in the world with around 300 facilities and counting. The Financial Times reports that Virginia’s data centers consumed at least 1.85 billion gallons of water in 2023, compared to 1.13 billion gallons in 2019. The rise of generative AI is driving the growth of data centers and increased demand for energy and water [11]. Throughout the U.S. and around the globe, data centers are located in water-stressed regions. This is becoming true for Virginia, too, as droughts are increasing and water withdrawals are exceeding supply [12].
On average, data centers’ traditional cooling methods use 300,000 gallons of water each day. Traditional methods involve using conventional air conditioning equipment, large fan systems, and smaller fans inside the equipment to pull in cooler air and reject warm air [4]. In addition to their high water usage and the chemical pollution caused by air conditioning coils, these methods are the least energy-efficient and increase operational costs, highlighting the need for more sustainable and efficient alternatives [4].
Data centers can use liquid or dry cooling methods. Dry, or air-cooling, methods lead to higher energy consumption, reduced performance, and increased operational costs. Therefore, the data center industry is shifting to liquid-cooling methods in data centers, according to the Association for Computer Operations Management’s “State of the Data Center Report” [5]. Liquid-cooling may be evaporative or direct water-cooling, which recycles loops of warm water to cool data center systems. The systems may use potable or non-potable water to absorb and carry away heat. Compared to dry cooling, liquid methods achieve better temperature control and reduced energy consumption [6]; however, evaporative cooling in particular uses millions of gallons of water each year [4]. In 2019, 57% of the water directly and indirectly used for cooling was from potable water [1]. Loudoun County’s data center potable water consumption has increased by 250% in the last 4 years, totaling 899 million gallons in 2023. Data center development expectations in other parts of the state are also triggering additional surface water withdrawal requests. [14] The method of utilizing non-drinking water sources for cooling purposes would help conserve potable water and reduce overall water usage in large data centers.
Data centers are often located near water bodies to source their water, and more than 10% of watersheds across the nation are already “water-stressed” watersheds or sub-basins [8]. More than 40% of the water scarcity footprint is attributed to data centers’ direct water consumption [2]. Yet, only a quarter of the total volume of water used by data centers result from onsite water use. This indicates that direct water consumption by data centers (which occurs close to where the data center is located) is skewed toward water stressed basins compared to their indirect water consumption. Recent drought conditions here in the Chesapeake region have shone a light on a growing problem of a fast growing demand on an increasingly stressed water supply [13].
In addition to the water used directly for cooling, water is also used indirectly to generate electricity needed to power the data center itself or or for wastewater treatment facilities servicing the data center. For example, it can take up to 11.86 gallons of water to produce just one kilowatt hour(kWh) of power through the use of fossil fuels [2]. That’s enough water to fill a bathtub up halfway! The 2024 State of the Data Center Report states that over 73% of data centers plan to utilize renewable energy, particularly solar (59%) and wind (28%)[5]. Renewable energy sources would help decrease indirect water usage.
Although it is known that data centers use a lot of water, there are many gaps in the data and information currently available. This is partially because data center water usage is not as well studied as energy usage. Data centers are not always required to document their direct water consumption [7]. Less than one third of data centers measure their water consumption. And among the ones that do, it is suggested that water consumption in data centers is higher than what is reported [7]. Clearly, better water consumption estimates and models are needed, especially to track water usage across different U.S. regions varying in humidity, irrigation, and watershed abundance levels [7].
Holding industry accountable
Big data is a multi-billion dollar industry, yet these facilities are not being made to pay for their impacts on our watersheds, nor are the laws sufficient to require them to reduce the impact. There are also numerous issues with local permitting processes granting these facilities the right to use our waters and allowing communities to respond to proposed new developments. There is a great need for further studies to quantify the impact of these facilities, before they are constructed, as well as monitoring the ongoing impacts to our waters.
In Virginia, the Joint Legislative Audit & Review Commission (JLARC) passed a resolution in December 2023 to study data centers and better understand the wide-ranging impacts. But we already know that local review of data center developments is inadequate to evaluate widespread impacts on the grid, electric ratepayers, water resources, parks, air quality, and emissions [14]. Meanwhile, Maryland is beginning to catch up with Virginia’s explosive data center growth. A major data center project, the first in Maryland, is already underway in Frederick County. In Maryland, a new data center stakeholder group has been set up to discuss issues related to the industry. Senator Ben Cardin stated a new process is required “to address the short and long-term cumulative impacts of each additional data center on land use, energy and water consumption, and impact on Maryland’s climate and energy efficiency goals [9].”
Waterkeepers in the Chesapeake region are working with communities, researchers and nonprofit partners to urge local and state lawmakers to not allow this global industry consisting of the largest companies in the world devastate our local rivers and aquifers. Current regulations and oversight are woefully inadequate. We urge residents to learn more and get engaged to demand that lawmakers stop using our tax dollars to subsidize the industry, ask the hard questions of where will the vast quantities of water and energy come from, and safeguard our public water resources that belong to all of us and our future generations.
References
[1] D. Mytton, Data center water consumption, NPJ Clean Water 4 (2021). https://doi.org/10.1038/s41545-021-00101-w.
[2] M.A.B. Siddik, A. Shehabi, L. Marston, The environmental footprint of data centers in the United States, Environmental Research Letters 16 (2021). https://doi.org/10.1088/1748-9326/abfba1.
[3] N. Lei, J. Lu, Z. Cheng, Z. Cao, A. Shehabi, E. Masanet, Geospatial assessment of water footprints for hyperscale data centers in the United States, J. Phys.: Conf. Ser. 2600 (2023) 172003. https://doi.org/10.1088/1742-6596/2600/17/172003.
[4] Lenovo, 3BL Blogs – 3BL Media: The Worlds AI Generators: Rethinking Water Usage in Data Centers To Build a More Sustainable Future, Newstex Global Business Blogs (2024). https://www.proquest.com/docview/2988505363/citation/92AB325AAB9D4913PQ/1
[5] S. MacGregor, Data Center Construction Trends: Build Fast, Build Smart, Data Center Knowledge | News and Analysis for the Data Center Industry (2024). https://www.datacenterknowledge.com/buildconstruction/data-center-construction-trends-build-fast-build-smart.
[6] Industry Voice by Staubli, The AI Boom, the Rise of Liquid Cooling, and the Universal Quick Disconnect, Data Center Knowledge | News and Analysis for the Data Center Industry (2024). https://www.datacenterknowledge.com/industry-perspectives/ai-boom-rise-liquid-cooling-and-universal-quick-disconnect.
[7] M. McCarthy, C. Brogan, J. Shortridge, R. Burgholzer, J. Kleiner, D. Scott, Estimating Facility-Level Monthly Water Consumption of Commercial, Industrial, Municipal, and Thermoelectric Users in Virginia, JAWRA Journal of the American Water Resources Association 58 (2022) 1358–1376. https://doi.org/10.1111/1752-1688.13037.
[8] Lindsey, Rebecca, NOAA Climate, Climate change to increase water stress in many parts of U.S., https://www.climate.gov/news-features/featured-images/climate-change-increase-water-stress-many-parts-us. (2013)
[9] Kurtz, Josh, Maryland Matters, Cardin expresses concerns about data centers in Md., power line project, https://marylandmatters.org/briefs/cardin-expresses-concerns-about-data-centers-in-md-power-line-project/. (2024)
[10] Osaka, Shannon, The Washington Post, A new front in the water wars: Your internet use, https://www.washingtonpost.com/climate-environment/2023/04/25/data-centers-drought-water-use/ (2023)
[11] Verma, Pranshu and Shelly Tan, The Washington Post, A bottle of water per email: the hidden environmental costs of using AI chatbots, https://www.washingtonpost.com/technology/2024/09/18/energy-ai-use-electricity-water-data-centers/ (2024)
[12] Maruccia, Alfonso, TechSpot, Virginia data centers consume too much water, and things will only get worse with AI, https://www.techspot.com/news/104358-virginia-data-centers-consume-much-water-things-only.html (2024)
[13] Council of Governments declares regional Drought Watch, https://www.mwcog.org/newsroom/2024/07/29/council-of-governments-declares-regional-drought-watch/ (July 2024)
[14] Virginia Conservation Network, Responsible Data Center Development, https://vcnva.org/agenda-item/responsible-data-center-development/ (2024)
EN 
ES