A recent study estimates that the current rate of Artificial Intelligence (AI) growth in the US could release between 24 and 44 million metric tonnes of carbon dioxide annually—an emission level comparable to adding 50 to 100 lakh (5 to 10 million) cars to US roads.
Carbon and water footprints
The research, published in the journal Nature Sustainability, also projects that the growth in AI usage will consume 731 to 1,125 million cubic metres of water every year. This water usage is estimated to be equal to the annual household consumption of 60 to 100 lakh (6 to 10 million) Americans.
Researchers from Cornell University suggest that the cumulative toll of AI use on the environment could prevent the industry from achieving its net-zero emissions targets.
"Artificial intelligence is changing every sector of society, but its rapid growth comes with a real footprint in energy, water, and carbon," said lead researcher Fengqi You, a professor of energy systems engineering at Cornell University’s College of Engineering.
Professor You explained the motivation for the study: "Our study is built to answer a simple question: Given the magnitude of the AI computing boom, what environmental trajectory will it take? And more importantly, what choices steer it toward sustainability"?
Data collection and projections
To understand the AI industry's expansion, the researchers began by compiling data related to various aspects of the financial, marketing, and manufacturing sectors. This data was combined with location-specific information on power systems and resource consumption, examining their relation to changes in climate.
The authors show that the deployment of AI servers across the US could generate an annual water footprint ranging from 731 to 1,125 million cubic metres and additional annual carbon emissions from 24 to 44 million metric tonnes of carbon dioxide equivalent between 2024 and 2030, depending on the scale of expansion.
Strategies for sustainability
Charting the environmental impacts of AI use state-by-state, the team discovered that data centres are increasingly being built in water-scarce regions, such as Nevada and Arizona.
The researchers found that strategically locating data facilities in regions with lower water stress and improving the cooling efficiency of data centres could slash AI's water demands by approximately 52 per cent. Furthermore, by combining this with best grid and operational practices, AI's water demands could be cut by as much as 86 per cent.
The researchers proposed smart siting, faster grid decarbonisation, and operational efficiency as key measures to cut carbon emissions due to AI use by 73 per cent and water demands by 86 per cent.
However, the authors noted a significant constraint: "Even as best practices may reduce emissions and water footprints by up to 73 per cent and 86 per cent, respectively, their effectiveness is constrained by current (clean) energy infrastructure limitations".
ALSO READ: WhatsApp to reportedly allow users to use same usernames as their Instagram and Facebook accounts
