Data centers’ rapid growth is driven by an increasing demand for AI-powered applications.
By 2030, U.S. data center electricity demand is forecasted to account for 9% of the total electricity consumed, placing a burden on the electricity and water infrastructures.
New cooling technologies and power management help reduce electricity consumption.
Data centers support infrastructure for data processing, housing computers and servers that store and process data for online platforms, streaming, smart infrastructure, autonomous vehicles, and AI applications (Coucil of State Governments Midwest 2024). Increased demand for data from emerging technologies, such as AI has led to rapid growth in data center workloads (Koot & Wijnhoven 2021; Masanet et al. 2020; Siddik et al 2021).
Data centers are grouped into small-scale and large-scale centers. Small-scale centers serve localized needs of small businesses, government, and larger corporations (Electric Power Research Institute (EPRI) 2024). Large-scale data centers, which include hyperscale data centers, serve operations such as multiple businesses or entire industries (EPRI 2024). In the U.S. there are currently 3,059 data centers in 50 states and D.C. (Data Center Map). In March 2024, Google selected Kansas City, MO for the location of a new data center (Butler 2024).
AI models, such as ChatGPT, are significantly more energy-intensive than past data center applications, requiring up to 10 times the electricity of traditional Google searches (Goldman Sachs 2024). Data centers are projected to consume up to 9% of U.S. electricity by 2030, from an estimated 4% currently (EPRI 2024; Figure 1). Electricity consumption estimates depend on AI industry trends, Internet traffic, storage demand, and AI model development.
Figure 1. Projections of potential electricity consumption by U.S. data centers: 2023–2030. Projections are based on low, moderate, high, and higher growth scenarios. The blue line illustrates increase from 2000 to 2010. Between 2010 and 2020 the rate of load growth leveled off. Figure taken from EPRI (2024).
Data center energy consumption is primarily driven by IT equipment (45%), cooling systems (40%), and auxiliary components (15%) (Ahmed et al. 2021; Shehabi et al. 2016). Cooling plays a critical role in maintaining hardware performance and longevity (Koot & Wijnhoven 2021).
Data centers rely heavily on water for continuous cooling and electricity needs, raising concerns about overburdening electrical and water infrastructure, as well as land resources (Al Kez et al. 2022; Mytton 2021). Approximately 20% of U.S. data centers rely on watersheds facing moderate to high stress from drought and other factors (Siddik et al. 2021). About 16% of 122 data center operators have disclosed plans for managing water-related risks such as cutting freshwater use, water reduction goals and deadlines, and tracking their water consumption (Johnson & Molnar 2022). The European Union (EU) requires data centers to report data on energy and water use (EU 2023).
Shifting from small-scale data centers to energy-efficient hyperscale data centers has reduced infrastructure energy use (Masanet et al. 2020).
Since the average workload in many data centers is typically around 30% of peak capacity, using power management techniques to power down idle devices and ensure only necessary ones are active, minimizes energy waste and helps reduce energy consumption (Shuja et al. 2012). Keeping an inventory of server hardware and applications, and linking applications to their physical servers help identify unused servers (DOE 2017). Energy efficiency standards for IT equipment (storage, computers, servers, monitors) provide data center operators with access to IT devices that offer greater energy efficiency (DOE 2017; Masanet et al. 2020).
Virtualization is a method for managing power and resources. It allows one single physical server to run multiple virtual computers, each handling its own applications (Shuja et al. 2012).
New technologies in cooling systems can allow data centers to manage future energy demands by reducing the energy-intensive nature of traditional cooling methods (DOE 2024; Masanet et al. 2020; National Renewable Energy Laboratory 2023). For example, immersion cooling could reduce cooling costs by up to 90% (Koot & Wijnhoven 2021).
Using renewables sources and investing in renewable-energy plants and technologies are a strategy employed by hyperscale data centers to support their expansion (McKinsey & Company 2023). Renewable energy reduces reliance on power grids and eases the pressure on electricity infrastructure (EPRI 2024; Sheme et al. 2018).
References
Ahmed KMU, Bollen MHJ, Alvarez M (2021). A Review of Data Centers Energy Consumption and Reliability Modeling. IEEE Access, 9: 152536-152563. https://ieeexplore.ieee.org/abstract/document/9599719
Al Kez D, Foley AM, Laverty D, Del Rio DF, Sovacool B (2022) Exploring the sustainability challenges facing digitalization and internet data centers. Journal of Cleaner Production, 15(371):133633. https://doi.org/10.1016/j.jclepro.2022.133633
Butler G (2024) Google to build $1bn data center in Kansas City, Missouri [Internet]. The Data Center Construction Channel. Link
Council of State Governments (CSG) Midwest (2024). Big data centers, big rewards for states? https://csgmidwest.org/2024/11/18/big-data-centers-big-rewards/
Data Center Map (n.a.) USA Data Centers https://www.datacentermap.com/usa/
Electric Power Research Institute (EPRI) (2024) Powering Intelligence: Analyzing Artificial Intelligence and Data Center Energy Consumption. https://www.epri.com/research/products/000000003002028905
European Union (EU) (2023) Energy Efficient Directive. Directive - 2023/1791 - EN - EUR-Lex
Goldman Sachs (2024) AI/data centers’ global power surge and the Sustainability impact. Link
Johnson E, Molnar K (2022) ESG Risks Affecting Data Centers: Why Water Resource Use Matters to Investors. Sustainalytics. Link
Koot M, Wijnhoven F (2021) Usage impact on data center electricity needs: A system dynamic forecasting model. Applied Energy, 291(116798). https://www.sciencedirect.com/science/article/pii/S0306261921003019
Masanet E, Shehabi A, Lei N, Smith S, Koomey J (2020) Recalibrating global data center energy-use estimates. Science, 367(6481):984-6. https://www.science.org/doi/abs/10.1126/science.aba3758
McKinsey & Company (2023) Investing in the rising data center economy. Link
Mytton D (2021) Data centre water consumption, npj Clean Water, 4(11): 1-11. https://doi.org/10.1038/s41545-021-00101-w
National Energy Renewable Laboratory (2023) Sandia National Laboratories’ Holistic Data Center Design Integrates Energy- and Water Efficiency, Flexibility, and Resilience. https://www.nrel.gov/docs/fy23osti/84310.pdf
Shehabi A, Smith S, Sartor D, Brown R, Herrlin M, Koomey J, Masanet E, et al. (2016) United States Data Center Energy Usage Report. Lawrence Berkeley National Laboratory. https://escholarship.org/uc/item/84p772fc
Sheme E, Holmbacka S, Lafond S, Lučanin D, Frashëri N (2018) Feasibility of using renewable energy to supply data centers in 60 north latitude. Sustainable computing: Informatics and systems, 1(17):96-106. https://www.sciencedirect.com/science/article/abs/pii/S2210537916300907
Siddik MAB, Shehabi A, Marston L (2021) The environmental footprint of data centers in the United States. Environmental Research Letters, 16(6): 064017. DOI 10.1088/1748-9326/abfba1
Shuja J, Madani SA, Bilal K, Hayat K, Khan SU, Sarwar S (2012) Energy-efficient data centers. Computing, 94(12):973-94. https://link.springer.com/article/10.1007/s00607-012-0211-2
U.S. Department of Energy (DOE) (2017) Small Data Centers, Big Energy Savings: An Introduction for Owners and Operators. Link
U.S. Department of Energy - Federal Energy Management Program (2024) Best Practices for Energy-Efficiency Data Center Design. https://www.energy.gov/sites/default/files/2024-07/best-practice-guide-data-center-design_0.pdf
