Featured · Common ground meets affordability
Are data centers raising your electric bill?
Yes — but not the way you'd expect. Living near one barely moves your bill. The cost hits everyone across a whole regional grid at once — about $9.3 billion this cycle in the Mid-Atlantic/Midwest, spread across 67 million people in 13 states, whether or not a data center is anywhere near you. A rural family that's never seen one is helping pay for it. The bills are rising now, and set to rise more — what's still a projection is exactly how much.
AI data centers are now roughly half of all new U.S. electricity demand. Residential electricity prices are up ~27% since 2019 and rose 11.5% in 2025 alone — and utilities asked for $29 billion in rate increases in just the first half of 2025, double the year before.
Not all of that is data centers — but they're a fast-growing piece, and the one almost nobody voted for.
What Americans think
PERCEPTIONMost don't want one nearby — and it's bipartisan
71% oppose71% oppose an AI data center being built in their area (48% strongly); just 27% favor one. It splits both parties — about 75% of Democrats and 63% of Republicans — and barely moves by age, race, education, income, or city-vs-rural. Opposition is even higher than for a nuclear power plant (53%).
Source: Gallup, March 2–18 2026 (n=1,000). Regional opposition: Midwest 76%, South 75%, East 68%, West 63%
What people think they do — good and bad
Share of U.S. adults saying data centers are mostly good vs. mostly bad for each:
Home energy costs6% good · 38% bad
The environment4% good · 39% bad
Quality of life nearby6% good · 30% bad
Local jobs25% good · 15% bad
Local tax revenue23% good · 12% bad
People credit jobs and tax revenue but blame energy costs, the environment, and quality of life. Two caveats: more than 20% answer "not sure" on each, and here the split is partisan — Democrats are more negative (50% say bad for the environment vs. 31% of Republicans). The more someone has heard about data centers, the more negative they get.
Source: Pew Research, 8,512 U.S. adults, Jan 20–26 2026
What the data shows: your bill
Believing something doesn't make it true — so here's the measured record, separated from what's still a projection.
The cost is regional, and you can't opt out
MEASURED OUTCOMEThe price to guarantee enough power in the PJM grid (the "capacity" auction) went from $28.92 to $269.92 per MW-day in one year — an ~833% jump — and hit the price cap the next two years ($329 and $333). PJM's independent market monitor pinned 63% of that increase (~$9.3 billion) on data-center demand. Crucially, that cost is spread across all 67 million PJM customers in 13 states + DC, not just where the data centers sit. It's already on bills: Pepco customers in DC are paying about +$21/month, roughly half of it from this.
Source: PJM Base Residual Auctions; Monitoring Analytics (PJM's independent market monitor); IEEFA; utility rate filings
Living next to one barely changes your bill — and that's the problem
MEASURED OUTCOMEA rigorous study (Rutgers/Bloustein, June 2026; 22,834 places, 24 states, 2,277 data centers) found ZIP codes with a data center had bills only about $2.09/month higher — "borderline" statistically — and at the utility level where rates are actually set, the effect fell to $0.35/month and wasn't significant. That's not reassurance. It means the cost isn't local — it's socialized across the whole region through the capacity market. A rural Pennsylvania family that has never seen a data center is helping subsidize one. The promised benefits are concentrated; the bill is diffused.
Source: Rutgers/Bloustein (NJ State Policy Lab); Lawrence Berkeley National Laboratory
Where it's headed
PROJECTIONIf the cost of serving all the planned data centers keeps getting spread across everyone's bills, one estimate (NRDC) projects the average PJM household could pay on the order of $70/month more by 2028. Treat that as a projection, not a measured fact — it hinges on choices regulators haven't finalized about who pays for new capacity. Electricity prices overall are projected up to 40% higher by 2030 vs. 2025.
Source: NRDC; EESI (projection / advocacy estimates)
What the data shows: the jobs & the tax breaks
The jobs rarely pay for the breaks
MEASURED OUTCOMEA finished data center runs on 15–100 permanent workers — an Amazon warehouse employs 1,000–1,500. Yet the subsidies are enormous: states have handed out incentives worth more than $2 million per permanent job. In Orangeburg, NY, one project got $77 million in tax breaks for exactly one permanent job. Virginia's data-center sales-tax exemption alone cost $1.6 billion in FY2025, and data centers took 53% of the state's economic-development incentives since 2021 (~$2.7B). Brookings' verdict: the incentives are poorly targeted and "may simply be subsidizing investments that would have happened anyway."
Source: Brookings (May 2026, ~770 facilities); Bloomberg Tax / NY Focus; CSG South / Virginia JLARC; LaborWise; MultiState
The pattern: concentrated benefit, diffused cost
A few companies get the tax breaks and the AI infrastructure; 67 million people across 13 states pick up the tab on their power bills. The profits are private; the costs are socialized — structurally the same pattern as trickle-down. That's why opposition is bipartisan: even President Trump posted that tech companies building data centers should "pay their own way." It isn't a left/right fight — it's a who-pays fight.
WHAT WE CAN PROVE VS. WHAT WE SUSPECT
- Proven: the PJM capacity spike is real and ~63% data-center-driven; it's already on bills regionally (Pepco +$21/mo).
- Proven, and counter-intuitive: the effect is regional, not local — your ZIP having a data center barely changes your bill.
- Proven: the jobs are few and the subsidies often exceed any plausible payoff.
- Projected, not yet measured: the big per-household numbers ($70/mo by 2028) depend on regulatory choices still to come.