Canton, a town in North Carolina, has blocked the construction of data centres through a one-year moratorium passed by the town’s Board of Aldermen. The move came in response to data centre firms attempting to acquire the site of an old paper mill that has been closed for years.
Residents of the town fear that data centres, crypto mining, and server farms will consume enormous amounts of water and electricity, which can strain local supplies. Broader environmental concerns, including water and noise pollution, were also cited as the reasons for the ban in the meeting convened on February 11, where more than 100 residents attended.
By passing the moratorium unanimously, Canton joined the trend of North Carolina towns rejecting or pausing data centres. Areas including Waynesville, Tarboro, and Matthews have also scrapped data centre plans, citing similar concerns.
Key Concerns on Data Centres Amid the AI Boom
Along with the aggressive growth of data centres amid the AI explosion, their massive power and water consumption are becoming a growing concern. The U.S. data centres consumed 4-4.4% of the national electricity supply in 2023. This is ~176 TWh, somewhat similar to the power consumed by a small country like Ireland. The U.S. Department of Energy (DOE) projects that data centres could consume up to 12% of total electricity in 2028.
Apart from straining the power supply, data centres directly consumed 17 billion gallons of water in 2023, a figure that has continued to rise in the years that followed. Moreover, the wastewater discharge and noise pollution are serious environmental concerns that nudge most towns towards banning them.
Town’s Debunking Economic Myths
Data centres are often touted as economic powerhouses that resuscitate a Town’s economy through job creation and tax windfalls. These claims often fail when met with reality, and the narrative of framing data centres as “economic saviours” is being rejected as the evidence suggests otherwise.
The job creation through data centres is short-term; it peaks during construction, and falls to ~100 employees per facility once it’s functional. These permanent employees are likely to be skilled and non-local workers. Moreover, most states offer tax breaks for data centres to attract new industries that can pay back in the long term. This distributes the burden of the increased power charge on the locals, for a projected tax windfall that may or may not come.
In short, data centres fail to pass the cost-benefit analysis of most towns since what they bring to the table does not justify the concomitant negative effects. While data centres are massive investments with high demand, small towns are becoming less likely to buy into them for projected long-term revenue gains.
Outlook and Implications
AI hyperscalers are clearly not easing up on their expansion since hardware providers like Nvidia are struggling to meet the AI demand. The demand has grown intense enough for memory chip companies to shift their focus from consumer products to catering to the AI infrastructure.
The hype experienced by AI data centre stocks in the share market is less likely to carry on to towns where companies plan to build infrastructure. The current modus operandi of hyperscalers can benefit investors, but fails to trickle down to the locals. This incentive mismatch is likely to continue muddying AI infra prospects in small towns. These permanent employees are likely to be skilled and non-local workers. Moreover, most states offer tax breaks for data centres to attract new industries that can pay back in the long term. This distributes the burden of the increased power charge on the locals, for a projected tax windfall that may or may not come.
In short, data centres fail to pass the cost-benefit analysis of most towns since what they bring to the table does not justify the concomitant negative effects. While data centres are massive investments with high demand, small towns are becoming less likely to buy into them for projected long-term revenue gains.
Outlook and Implications
AI hyperscalers are clearly not easing up on their expansion since hardware providers like Nvidia are struggling to meet the AI demand. The demand has grown intense enough for memory chip companies to shift their focus from consumer products to catering for the AI infrastructure.
The hype experienced by AI data centre stocks in the share market is less likely to carry on to towns where companies plan to build infrastructure. The current modus operandi of hyperscalers can benefit investors, but fails to trickle down to the locals. This incentive mismatch is likely to continue muddying AI infra prospects in small towns.
