started · updated
AI expansion drives surge in data center energy and water demand
The rapid expansion of artificial intelligence is driving unprecedented demand for data center infrastructure, leading to significant environmental challenges regarding energy, water, and waste.
Global data center electricity consumption is projected to reach approximately 448 terawatt-hours in 2025, with AI-specific consumption potentially hitting 945 terawatt-hours annually by 2030. In the United States, data centers could account for up to 11.8% of total electricity consumption by 2030. This surge is compounded by cooling requirements, which can account for roughly 40% of a data center's energy use, a figure that may rise during extreme heat events.
Beyond energy, AI infrastructure poses risks to water resources and land use. Large-scale data centers can consume millions of gallons of water daily for cooling. Furthermore, AI infrastructure is expected to generate up to 2.5 million tons of electronic waste annually by 2030.
To mitigate these impacts, companies are exploring alternative solutions. Microsoft previously tested underwater data centers near the Orkney Islands, finding that submerged servers experienced failure rates approximately eight times lower than terrestrial counterparts. However, experts note that while such innovations may improve cooling efficiency and reduce freshwater use, they do not eliminate broader concerns regarding carbon emissions and marine environmental impacts.
Entities
Andrew Sobko · Argentum AI · Google · International Energy Agency · Lawrence Berkeley National Laboratory · Microsoft · Nvidia · OpenAI · United Nations · United Nations University Institute for Water, Environment and Health · United States · Vertiv
Claims
What the coverage asserts, and how many sources carry each claim.
- [○ 1 SOURCE] Argentum AI is expanding its workforce to include specialists in power engineering, cloud architecture, and GPU infrastructure. www.webdisclosure.com
- [○ 1 SOURCE] Google aims to become 'water positive' by 2030. jornalempresasenegocios.com.br
- [○ 1 SOURCE] AI data center operators expect to secure necessary power supplies within two to three years.
- [● 3 SOURCES] Water used by AI data centers is expected to equal the everyday water needs of 1.3 billion people by 2030. knowridge.com · www.philstockworld.com · eveningreport.nz
- [○ 1 SOURCE] AI data centers and semiconductor clusters in South Korea are projected to require 38.4GW of additional power, totaling 260TWh annually. zdnet.co.kr
- [○ 1 SOURCE] Microsoft reported a 25% increase in greenhouse gas emissions compared to the previous year. jornalempresasenegocios.com.br
- [○ 1 SOURCE] In 2025, major technology companies invested over US$ 400 billion in physical infrastructure, primarily for data centers. aepet.org.br
- [○ 1 SOURCE] Argentum AI is expanding its global AI Factory platform and leadership team to support hyperscale AI infrastructure demand. www.webdisclosure.com
- [○ 1 SOURCE] A median text query to Google's Gemini uses 0.24 watt-hours of electricity and 0.26 milliliters of water. www.green.it
- [● 4 SOURCES] Data center electricity consumption is projected to reach approximately 945 TWh by 2030. www.green.it · www.truenumbers.it · www.digitaljournal.com · marketbusinessnews.com
- [○ 1 SOURCE] Data center electricity consumption has grown at an annual rate of 12% since 2017. www.truenumbers.it
- [○ 1 SOURCE] Cooling accounts for roughly 40% of data center energy consumption and may increase by 20–30% during extreme heat. bizblog.spidersweb.pl