A data center's coming and I'm on a well — do I drill deeper or add storage?

Does the data center even drink from my aquifer?

The comments all say the same thing: your well's gonna dry up. So I went looking for who it actually happened to. In Newton County, Georgia, a couple named Beverly and Jeff Morris live about a thousand feet from a Meta data center. Their taps went to mud. The dishwasher, the ice maker, the toilet — all of it quit on them.

But here's the part that stopped me cold. That data center wasn't even running yet. It hadn't pulled a single drop of groundwater. The damage lined up with the construction — the tree clearing, the blasting, the digging. One explanation is that all that shaking dropped the water table enough that their pump started drinking from the bottom of the glass, where the grit settles.

And a lot of these places don't tap your aquifer at all once they open. Plenty run on city water, or a closed-loop cooling setup that recycles the same water instead of guzzling fresh. So the first question isn't how deep is my well. It's where is this thing getting its water, and when does the risk actually land — during the build, or after it's humming? Call the county. Read the permit. That answer is public record, and it points you at a completely different plan depending on what it says.

If it does plan to pump groundwater near you, the thing to understand is well interference. A big well and your well can reach into the same water. Picture two straws in one glass — when the big one pulls hard, your end drops too. The closer you sit and the shallower your well, the more you feel it.

The cheap move everyone skips: measure your own well first

Here's what I didn't know. If your water goes bad a year from now and you want someone to pay for it, you have to prove what your water was like before. No baseline, no case. And almost nobody has a baseline.

A baseline test is just a well contractor coming out and writing down three things: how deep the water sits when nothing's pumping, how many gallons a minute your well actually makes, and what's in the water right now from a lab. Date it. File it somewhere you won't lose it.

Minnesota is actually trying to make the data center companies pay to test every well within two miles before they break ground. Most states don't require anything like that yet. So in most of the country, if you want that number on paper, you're the one who has to go get it.

The second reason to do it has nothing to do with lawyers. That same test tells you whether you even have a problem. Maybe your well's a monster with water to spare and you can stop losing sleep. Maybe it's already marginal and you'd have wanted to know regardless of any data center. You can't choose between drilling and storage until you know which one you are.

So — drill deeper, or add storage?

This is the real question, so here's what the trade actually is.

Storage is the cheap, boring answer, and it works better than it sounds. You put a tank in the basement or the garage, and your slow well fills it overnight while nobody's running water. A well that only trickles a gallon a minute still quietly makes around 1,440 gallons a day. Most houses don't use close to that. The tank just smooths out the busy mornings so you're not fighting the shower for pressure.

Drilling deeper is the expensive answer, and it's a gamble. Deeper can mean more water — or it can mean the same water with a bigger, pricier pump to lift it, or worse water you suddenly have to filter. In hard rock, going down farther or fracturing the well often doesn't buy much, because the new cracks still have to be fed by stone that barely moves water at all.

The one thing I couldn't pin down from my kitchen table is the piece that decides it: your local geology. That's what a driller who knows your specific area has to answer, and it's worth the call before you spend a dime either way. So the order that actually makes sense is: find out the data center's water plan, get your baseline, then pick storage or depth. Do it backwards and you can spend six grand solving the wrong problem.

Why I keep pulling on this thread

I started this wondering about one well and ended up three clicks deep in cones of depression, a congresswoman holding up jars of brown water at a hearing, and a retiree in Indiana who sold the house she'd lived in for 26 years after an Amazon build dried up her well. Every one of these stories opens with something that sounds dull — cooling, water tables, a permit filing — and ends at somebody's kitchen tap.

That's the whole reason Byte Bungalow exists. We take the stuff humming behind the chatbot — the power, the water, the gas turbines, the substations nobody notices until one shows up in the backyard — and follow it until it's about a real person. If that's your kind of rabbit hole, the full version of this one is over on our YouTube channel. Come watch it, and hit subscribe so the next thread lands in your feed.

Common questions

Keep reading

Sources

• sanjuandailystar.com — Beverly & Jeff Morris, Mansfield/Newton County GA, home ~1,000 ft from Meta data center; taps ran dry within months of construction; dishwasher/ice maker/washing machine/toilet sto
• ajc.com — Sept 2026: Georgia families sued Meta accusing the Newton County data center of tainting their water (confirms active lawsuit / timeframe).
• ppc.land — $750M Meta data center, Newton County; multiple households reported well failures, sediment buildup, deteriorating quality by completion; EPA figure: US data centers consumed 17.4B
• blog.andymasley.com — Key assumption-break: the Georgia well damage occurred during CONSTRUCTION; the data center "hadn't begun operations and doesn't draw groundwater"; theory that turbid water = water
• blog.se.com — Closed-loop systems circulate coolant without continuous evaporation, use negligible water vs evaporative; evaporative uses more water/less energy, closed-loop more energy/less wat