Data centers, urban heat, and AI growth: CT experts weigh in, including pushback
AI growth is driving significant data center expansion, leading to increased community pushback due to environmental concerns such as urban heat, water consumption, and electricity grid strain. Experts from Yale weigh in on the complex trade-offs, highlighting the need for careful policy and sustainable solutions to mitigate impacts. Cities like Denver, Minneapolis, and Seattle are considering municipal restrictions.
The rapid expansion of data centers, primarily fueled by the demands of artificial intelligence, is encountering growing opposition from communities concerned about environmental and economic impacts. Experts from the Yale School of the Environment are examining the complexities of this trend, with cities like Denver, Minneapolis, and Seattle already considering municipal restrictions on data center development.
Karen Seto, Director of the Hixon Center for Urban Sustainability, explains that nearly all electricity consumed by data centers converts to waste heat, requiring continuous cooling. Research in Phoenix, Arizona, indicates that a large data center can increase surrounding land surface temperatures by as much as 16°F and air temperatures by up to 4°F, exacerbating urban heat island effects. Daniel Esty, Professor of Environmental Law and Policy, frames the opposition as a "Rorschach Test" reflecting economic confidence, while also acknowledging the significant risks AI poses, including increased electricity demand that could spike greenhouse gas emissions if not met by clean energy.
Kenneth Gillingham, Professor of Environmental and Energy Economics, emphasizes that the race to train better AI models is the main driver of data center growth, consuming massive amounts of electricity and water. Some large data centers are projected to use over 3 GW of peak electricity, raising anxieties about rising electricity prices and reliance on natural gas. Yuan Yao, Professor of Industrial Ecology and Sustainable Systems, highlights the context-dependent nature of environmental impacts, noting that carbon footprints vary significantly based on the local energy grid and water use is a major concern in water-stressed regions. Yao's group is developing sustainability metrics to help compare impacts across various data center choices.
Overall, the article underscores the critical need for a clear-eyed approach to AI's benefits and costs, with a focus on policy-guided investments, improved energy efficiency, clean electricity procurement, and transparent reporting to ensure a sustainable future for communities amidst this technological boom.