Amazon says its datacenters used about 2.5B gallons of water last year

tcp_handshaker 15 points 18 comments June 14, 2026
www.theregister.com · View on Hacker News

Discussion Highlights (4 comments)

paytonjjones

There are a lot of dumb political discourses but this is one of the dumbest in years. It's puzzling to me how water got latched onto instead of, say, energy or rare earths.

jrflo

What drives me nuts about this whole water consumption discourse, is no one ever asks what happens to the water when it's used- it gets evaporated! It re-enters the water cycle as pure water! It's not like it's getting destroyed or turned into some unusable sludge, it'll just rain back down again in a few weeks. Obviously, you still shouldn't build datacenters in areas that are locally water-stressed as that will add additional burden to the infrastructure.

h4kunamata

And yet the population have to ration it because there is no enough water. This math isn't "mathing".

MisterMower

People have difficulty visualizing large volumes of water. A square mile filled up to one foot with water is 208,544,914 gallons. 2.5 billion gallons of water would fill up this same square mile to only 12 feet. It’s the size of tiny lake. For context, Lake Meade has about 250 square miles of surface area and an average depth of 180 feet. The reservoir my city of half a million people gets its water from is 15 square miles and 20 feet deep. 2.5 billion gallons is maybe 5% of our reservoir’s total volume. This headline is textbook fearmongering. They are intentionally misrepresenting the magnitude of water usage to make you fear data centers. There are plenty of reasons to be wary of them, but water usage is not one of them. The authors of this article are either assuming you’re to dumb to do simple volume conversions or worse, too stupid to do them themselves. I suspect the latter since they use inconsistent units throughout the article. Liters per kWh, cubic meters, gallons, etc. You don’t hate journalists enough.

Semantic search powered by Rivestack pgvector
10,500 stories · 98,695 chunks indexed