- Data centers and their energy needs are causing pollution
- This pollution is causing knock-on costs for healthcare and the environment.
- A cost of up to $25 billion is theorized, but it could be higher
The rise of AI and its demand for data centers could cost the US economy $25 billion in environmental damage and cause serious health complications for those living in close proximity to data centers.
These figures come from a new working paper written by Nicholas Z. Muller, an economist at Carnegie Mellon University, and published by the National Bureau of Economic Research.
The paper surveyed about 2,800 operating data centers on their energy needs and examined the impact of additionally generated air pollutants and greenhouse gases on the environment, the health of American citizens, and their cost to the economy.
Article continues below.
Lung disease, heart conditions, premature mortality.
In the paper, Muller calculates the gross external damages (GED) figure of $25 billion by calculating not only the use of artificial intelligence within data centers (which he attributes a figure of $3.7 billion), but also the broader indirect costs of pollutants caused by generating the energy needed to power the data centers.
Muller’s article theorizes that the planned expansion of data center construction could increase associated damages by up to 85% in the short term.
These costs include the additional burden on healthcare facilities and the daily medical needs of local residents, which Muller attributes to the negative effects of exposure to the airborne pollutant PM2.5, which can cause lung disease, heart conditions and higher rates of premature mortality.
There are also broader social costs associated with pollutants and greenhouse gases that have not yet materialized, but are nonetheless being emitted by the additional energy sources being built or revived to fuel the boom in energy demand.
The Trump administration’s stance against renewable energy sources means that legislation has been introduced to restart a number of previously decommissioned fossil fuel power plants to meet additional demand and, in turn, are releasing more pollutants and greenhouse gases.
These external costs are not borne by the technology companies that finance the construction of data centers, who instead enjoy significant tax breaks for their investments in new construction projects, generating less tax revenue to finance the construction of the additional infrastructure necessary to power and transfer electricity to data centers.
The Trump administration introduced the Taxpayer Protection Pledge to tie the additional costs of energy production to technology companies, but the pledge is completely voluntary and has very little in the way of accountability, timelines or provisions. Some communities in the US are now experiencing skyrocketing energy costs due to the power demands of data centers.
There are also the problems with the employment that data centers provide. While initial construction can provide a boost in employment, the effect is largely temporary. Many data centers are automated and require little supervision to operate.
Muller’s paper estimates that the economic costs of environmental damage and related healthcare costs are small compared to the potential productivity gains produced by AI, but residents living in the vicinity of data centers may disagree with this assessment.
There is currently growing anti-data center sentiment across the United States, with numerous grassroots local resident groups opposing the construction of data centers, causing significant project delays and cancellations. There also appears to be an overall downward trend in opinion of AI: up to 71% of US citizens believe AI could cause permanent job loss, and almost half (47%) believe it will have a negative effect on humanity.
Through Fortune
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds.
