So far in 2026, almost half of all planned data centers in the US have been delayed or cancelled, with the barriers ranging from chip shortages to insufficient energy infrastructure.
But local opposition is quickly becoming a new force that developers have to reckon with, especially after a single movement in Box Elder county, Utah, gained global attention.
A single slogan, “People Over Profits”, has been seen on signs, banners, and placards throughout data center opposition groups across the country.
Energy consumption
Electricity price comparison site Electric Choice estimates that in 2023, over 4,500 data centers consumed around 176 TWh of electricity, or enough to power more than 16 million homes per year.
That was roughly 4.4% of US annual electricity consumption in 2023. In 2024, the Electric Power Research Institute estimated that AI data center consumption could be between 10-20% of US electricity consumption.
Now, in 2026, that number will only have grown, especially as hyperscale projects such as OpenAI’s Project Stargate begin construction. But in most cases, developers and investors are either building their own energy generation or investing in new infrastructure to support their usage.
But this is also causing issues. Many projects have turned to natural gas turbine generators that are both polluting and loud. Even for projects built in rural areas, the drone of generators can carry for miles. For data centers in areas closer to population centers, numerous residents have complained of an illness caused by the infrasound emitted by datacenters.
In an attempt to quickly address the lack of capacity for data centers, the Trump administration has rolled back environmental protections and set aside funds to bring shuttered coal and gas power plants back online. Great for investors and developers, not so great for Americans.
Energy prices in areas close to data centers have skyrocketed by as much as 267% as demand surges. For context, a JDPower report published in January 2026 found that more than one in five Americans cannot afford their electricity bills.
Water usage
There is a significant debate around the water usage of data centers. Critics have argued extensively over the amount of water a single AI prompt uses.
For example, scientists at the University of California, Riverside, have placed the water usage for a single 100-word prompt to be around 519 milliliters, or around a single bottle of water. On the other hand, OpenAI CEO Sam Altman has claimed that a single prompt uses “roughly one fifteenth of a teaspoon”. Altman did not provide a source for his claim.
But to understand how much water is actually ‘used’, we need to understand how data centers use water.
The water used to power and cool AI data centers isn’t consumed during use. Large amounts of it evaporate to exchange the heat generated, with the remainder being recycled. The sources of water include groundwater, non-potable water, or municipal water systems.
The issue lies in how much water data centers pull from water sources in order to operate. Estimates from the Environmental and Energy Institute place the daily water consumption of a large data center at 5 million gallons per day. This amounts to the water use for a town populated by 10,000 to 50,000 people.
The water doesn’t disappear, but it is taken away from the local ecosystem, which can worsen the effects of droughts or ecological loss in areas with water constraints. Moreover, the added waste water produced by data centers can overwhelm the capacity of local treatment centers designed to handle the output of local populations.
Local jobs
Data center development has been sold to many communities as an opportunity to create jobs and enhance both the local and national economy. This is true.
But the jobs aren’t necessarily immediately available for local people, and aren’t necessarily permanent. Constructing an AI datacenter, especially the large ones, requires specialized construction equipment and expertise that local construction firms likely don’t have access to. This means the construction contracts are handed out to large firms, some of which will be based out-of-county.
Once the facility is constructed, these jobs are no longer required. The data center does require staff, but a small facility will only have a dozen or so onsite employees, with larger sites reaching the triple digits.
Where the real job growth can be claimed, as a Brookings study has shown, is in the professional services required to keep a data center running, such as fiber installation, network operations centers, managed service providers, and IT contractors. Six years after a data center opens, the employment of information sector and professional services have grown by around 22% and 16% respectively.
So there is an obvious balancing act to be played between energy and water consumption, local resident health, environmental protection, and job growth. AI developers in competition with each other, and the investors behind developments, may indeed be putting profit over people.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.
