
January 30, 2025
Your smartphone may fit in your pocket, but it requires vast amounts of land, water, and electricity to function.
Mobile devices have condensed the computing power that once took up space on people’s office desks and, prior to that, filled entire floors of buildings. This digital downsizing of computers seemingly gives us the convenience of a technological genie in a bottle. However, if we look behind the scenes of our screens, we see there are vast amounts of resources required to make this tech possible.
Anything you do online involves retrieving information from data centers, which are large campuses of warehouse-sized buildings that contain racks and racks of computers. These specialized computers that store data are called servers. You couldn’t watch videos, get GPS directions, send emails, shop online, or do anything else on the internet without these extensive collections of servers. The use of cloud storage services like Google Drive or iCloud for storing photos and other files online also relies heavily on data centers.
There are thousands of data centers in the U.S.; the larger ones are as big as four combined football stadiums! Tech companies like Microsoft and Meta are hungry for the limited real estate to construct and maintain these expansive facilities. And these large buildings use up more than just land. If you’ve ever had a laptop resting on your lap for a long time, you know that computers can get hot. It takes a great deal of water to keep rows and rows of computer servers at the right temperature.
As AI (Artificial Intelligence) models like ChatGPT become more prevalent and integrated into different applications, even more resources are required. It isn't easy to assess exactly how much water and energy different AI models use. Still, it is estimated that generating a simple 100-word email using ChatGPT uses the equivalent of a bottle of water.
Besides water, electricity is needed to make all of this work. In terms of electrical power, generating one AI image takes about the same amount of electricity required to charge a smartphone from 0 to 100%. Developing and training one of these new AI models takes 1,300 megawatt hours of electricity, about as much electricity as 130 households use for an entire year!
While advancements in AI have made it more challenging for tech companies to meet their energy use and emissions goals, new algorithms and AI advancements may offer solutions to the problems this technology has exacerbated. As droughts and wildfires continue throughout the country, figuring out this issue has increased importance.
MCPL Library cardholders can access the articles published by different news publications through the databases found on the Library’s eMagazines and eNews resource page.
Brad Smith
Digital Innovation Department
Read Similar Blogs:
Animals and Nature
Business
Current Issues and Topics
Science and Technology
Technology