Water for Tomorrow started with a simple frustration. Every week another headline claimed that artificial intelligence was draining the country dry, and every week the numbers underneath those headlines fell apart on inspection. This site is an attempt to put the real figures in one place, show the math, link every source, and let you decide for yourself.
of total U.S. daily water consumption came from data centers in 2023.
more water is used by California almonds than by all U.S. data centers combined.
real U.S. data center clusters mapped with verified power, water, and cooling data.
of the numbers on this site are cited to a primary source you can open.
Water scarcity is a genuine problem in parts of the United States. The Colorado River basin is stressed, the Central Valley aquifer is in overdraft, and the Ogallala is shrinking. Those are real and they deserve serious attention.
The problem is that the loudest recent story about water has been about the wrong thing. A single email written with a chatbot does not drink a bottle of water. Data centers are not the reason your town is rationing. When the public conversation aims its alarm at a use that accounts for a fraction of a percent of the total, it burns energy that could have gone toward the uses that actually move the needle.
This site steelmans both sides. It gives the strongest honest case for why data centers are efficient, and it names the specific local places where they have caused real harm. The goal is not to defend any industry. The goal is proportion.
Every figure traces to a government dataset, a peer reviewed study, or a named researcher who published their math. USGS, USDA NASS, BEA, LBNL, and the work of Andy Masley, Brian Potter, and Tomas Pueyo.
Water that evaporates or leaves as product is counted differently from water that is withdrawn and returned. The site uses consumptive figures so comparisons are fair rather than flattering.
The Claim Ledger does not just label a claim true or false. It shows the calculation that produced the verdict, so you can check the work instead of trusting the label.
A number that is reassuring at the national scale can hide a real problem in one county. The map keeps those two truths visible at the same time rather than letting one erase the other.
If you cannot open the source, the number does not belong here.
Steelman, not strawman. The opposing view is presented at full strength.
When a figure is shown to be wrong, it gets fixed in the open, the way Brian Potter corrected his own earlier water estimate.
The tone stays calm even when the topic is charged. Alarm is a tool, not a default.
I build interactive explainers that turn contested topics into something you can poke at with your own hands. This one grew out of reading the same misleading water headline one too many times. If you find an error, or you have a source that sharpens a number, I want to hear about it. The whole point is to be corrected in public.
Independent research software. No ads, no paywalls, no corporate sponsors, no user tracking. Your contribution funds development, data, and hosting.
Pour five gallons, read the claim ledger, calculate your own footprint, and open the map. Everything is sourced and nothing is hidden.