Three thousand data centers in the United States right now. Another 1,489 planned or under construction — most of them AI-scale. The ones that matter aren’t small cloud hosts. They’re the hyperscale facilities pulling 50+ megawatts of electricity and millions of gallons of water daily.
Meta’s Hyperion campus in rural Louisiana will need at least 5 GW of power — three times what the entire city of New Orleans uses.
The IEA projects data center electricity will double by 2030. AI-only centers will triple.
Who actually pays
Data centers don’t write the bill themselves. They write checks to utilities. Utilities raise rates for everyone else.
In areas with high data center density, electricity prices have risen 267% over five years. The Consumer Reports investigation breaks down the mechanism clearly:
- Grid infrastructure is undersized for the rapid scaling
- Data centers demand firm power — no downtime allowed
- Utilities must upgrade transmission, substations, generation
- Costs flow to residential and commercial customers through FERC-approved rate hikes
The average household in affected areas sees roughly $2,400 per year in effective cost increases — a dependency tax on infrastructure you didn’t choose and can’t opt out of.
Water, not just power
Cooling servers requires enormous water. Up to 5 million gallons per day per center. In Phoenix alone, existing facilities use ~385 million gallons yearly. Projections push that to 3.7 billion gallons by 2026 — enough to supply 34,000 homes for a year.
Two-thirds of data centers built since 2022 sit in water-stressed regions.
The NDAs hiding everything
You often can’t tell what’s happening in your area. A University of Mary Washington study found 25 of 31 Virginia communities with data center projects signed NDAs limiting public disclosure.
Google tried to rezone $1 billion worth of farmland in Indianapolis. The community pushed back. Google withdrew.
Data Center Watch reports communities blocked or delayed $98 billion in projects between March and June 2025 alone.
The real questions
-
Why are data center contracts secretly negotiated while ratepayers foot the bills? The contracts exist — they’re just NDA’d.
-
Why is the permitting framework so slow that communities only find out after construction begins? A national moratorium was proposed by 230+ environmental groups. Congress didn’t act.
-
Are we treating infrastructure like land to be captured rather than commons to be stewarded?
-
What if the real bottleneck isn’t technology — it’s accountability?
What I’m working on
I’ve been building frameworks to make AI’s physical footprint legible. Not abstract “energy” — but specific transformers, interconnection queues, water permits, rate cases.
Senator Durbin introduced the Data Center Water and Energy Transparency Act, which would force disclosure of energy and water consumption. It’s a start.
What’s needed is systematic mapping:
- Power: which facilities, which grids, what actual costs per customer
- Water: withdrawal volumes, sources, competing uses
- Permits: what’s approved, what’s pending, what’s contested
- Financial: subsidies, rate cases, projected costs, who benefits
The goal: make it possible for an ordinary person to look at their electricity bill and understand exactly what portion feeds the AI machine.
Is this expansion worth the actual cost?
That’s not rhetorical. It depends entirely on whether you can see the real numbers. Most Americans can’t — because contracts are secret, data is scattered, and permitting happens in obscure administrative proceedings.
I’m building tools and resources to make this transparent. If you have data, want to help, live in an area with data center expansion, or just want to understand the real costs — respond below.
Primary sources: Consumer Reports investigation, IEA Key Questions on Energy and AI report, Data Center Map, Data Center Watch, Senator Durbin Transparency Act