Data Centers: Local Action

Data centers: where you can actually act

The environmental page makes the case that AI's global footprint is small and often misframed. But specific data centers cause real, local harms — to air, water, grids, and electricity bills — and local is exactly where an ordinary person has leverage. This is the map of who is doing what, attributed to the right entity, and the concrete lever you can pull where you live.

$64B $18B blocked + $46B delayed over two years, across 142 groups in 24 states — and a newer tally puts full-2025 opposition near $156B (Data Center Watch / 10a Labs; self-reported, LLM-assisted, publicly-disclosed values only). Showing up to a county board meeting is not symbolic — it is the thing that actually stops projects.

Where your leverage actually is

The discourse fixates on the lowest-leverage action — feeling guilty about your own queries. Individual habits barely move a system built on capex you don't control. Your leverage climbs steeply at the local civic level, where a single rezoning vote or utility-commission docket decides whether a gigawatt campus gets built next to you and who pays for its grid.

Your own useskip a query
Consumer choicewhich services
Local civiczoning · permits · rate cases
★ highest leverage$64B blocked
▲ This rung is where the outcomes are decided. Every community win in the map below — and the $64B in blocked & delayed projects — was fought and won here, at a county board, city council, or utility commission.
State policylegislature · PUC
rate design, siting law
Federal / globalslow, diffuse
high impact, low access

Bar width = a rough sense of an individual's realistic influence per unit effort, not a measured quantity. The gridlines are visual reference marks, not data. The point is the shape: the middle rung, not the bottom one, is where you change outcomes.

Is AI really why your electric bill is high?

This is the framing worth getting right. Nationally, for most people, data centers are not the main driver of their electricity bill — fuel prices, transmission upgrades, storm hardening, and inflation dominate. But in a few data-center-dense grids, there is a real, measured cost-shift onto ordinary ratepayers. Both things are true, and the news tends to collapse the second into a universal version of the first.

The headlines

"Data centers drive a 76% surge in PJM power prices" (Bloomberg, E&E). Virginia's own legislature (JLARC) estimates a typical Dominion customer could pay $14–$37 more per month by 2040. Harvard researchers warn utilities are "forcing ratepayers to fund discounted rates for data centers."

PJM / Monitoring Analytics · JLARC · Harvard EELP

What the 76% actually is

The 76% is a wholesale energy price ($136.53 vs $77.78/MWh) — driven mostly by fuel, demand and weather, not AI. The genuinely data-center-attributable figure the PJM monitor isolates is separate: capacity costs up +398% (vs +5% for transmission), ~$13B tied to data-center load in the last two auctions.

It's also localized (a PJM/Mid-Atlantic wholesale effect, not your retail bill) and, nationally, outweighed by other drivers — extreme weather, ~$1.4T of grid capex (up 27%/yr), and rising utility profits. Fortune's summary: data centers are "the scapegoat … the reality is more complicated."

The reporting collapses two different meters into one number. They measure different things — read each with its own label, not one against the other:

↑ 76% wholesale energy price
$77.78 → $136.53 per MWh
The number in the headline. Driven mostly by fuel, demand & weather.
this part is NOT AI
↑ 398% wholesale capacity price
~$13B tied to data-center load (vs +5% transmission)
The part that really is data centers — but wholesale & localized to PJM, and nationally outweighed by weather + ~$1.4T grid capex.
real — but NOT your retail bill

two different measures on two different meters · a bigger % here is not a bigger hit to your monthly bill

The honest civic message: in specific grids there's a real cost-shift worth contesting at your utility commission — and it's fixable. In July 2025 Ohio's PUCO approved an AEP Ohio tariff making 25 MW+ data centers pay for 85% of their reserved capacity for up to 12 years, which it found "safeguards other non-data-center customers from cost-shifting." That is the concrete ask. In most of the country, the "AI is why your bill is high" headline is louder than the effect on your own bill.

The map: data centers drawing complaints

Every entry is from a fetched source and attributed to the correct entity (operator vs. parent vs. developer). Allegations are labeled as such. Green = a community already won — proof the levers work.

How to actually act locally

1

Find the proceeding

Most data centers need a rezoning or a special-use / conditional-use permit — decided at a public hearing before your county board or city council. Ratepayer and grid fights happen at the state utility commission. Use the Data Center Watch directory to find the project and body near you.

2

Show up to the public comment period

Permits and rules have a formal public comment period; comments enter the official record, and in some cases the agency must respond. Speaking at the hearing — or organizing neighbors to — is the single highest-leverage act here.

3

Target the specific harm to the specific body

water → city water utility & annexation vote  ·  air → county health dept & Clean Air Act  ·  bills/grid → state PUC rate case  ·  land/noise → zoning & rezoning hearing

4

Ask for the cost-shift fix

At the utility commission, the concrete ask is a data-center rate class / minimum-take tariff so large users commit to and pay for the grid capacity they request — keeping system costs off residential bills.

Sources

Part of the AI Problems Index. This page catalogs documented complaints and their local levers; it is not legal advice, and pending allegations are labeled as such. Have a local case with a citable source? It belongs here.