Can underwater data centers ease AI's environmental toll?
This article explores the potential of underwater data centers, exemplified by a project in China and Microsoft's Project Natick, to mitigate the environmental impact of AI by reducing energy and water consumption. However, experts caution that more research is needed to fully understand their effects on marine environments, including thermal pollution and noise. The rapidly growing AI market faces significant environmental concerns regarding electricity and water use, making innovative solutions crucial.
The potential for underwater data centers to alleviate the environmental impact of artificial intelligence is being explored, with China operating a 24-megawatt facility off the coast of Shanghai. Built by HiCloud Technology in partnership with the Chinese government, this data center is wind-powered and reportedly reduces power consumption by 23%, and water and land use by 100% and 90% respectively.
This development follows Microsoft's Project Natick, which successfully tested a submerged data center off Scotland between 2018 and 2020, proving the concept's feasibility for reliability and sustainability. However, a Microsoft spokesperson confirmed to The National News Desk that the company currently has no operational underwater data centers but continues to use Project Natick for research.
The global AI market's rapid expansion, projected to reach nearly $5 trillion by 2033, raises significant environmental concerns. A United Nations University report highlighted that global data centers consumed an estimated 448 terawatt-hours of electricity last year and predicted that their water demands could equal that of 1.3 billion people by 2030.
Eric Masanet, a professor at UC Santa Barbara specializing in data centers and climate change mitigation, suggests it's too early to definitively say if underwater data centers offer a net environmental benefit due to limited empirical data. He notes that using surrounding water for