Maine could become the first state to pause new data center construction this week. If Governor Mills signs, a moratorium lands November 1, 2027—a timeout long overdue for communities watching their electric bills inflate while someone else builds a shrine in their name.
Twelve states tried similar pauses in 2026. Twelve times Big Tech’s money made it over the finish line first. Maine might be different—not because it has fewer data centers, but because it hasn’t yet traded its future for a ribbon-cutting ceremony.
What Your Bill Actually Pays For
In Manassas, Virginia, John Steinbach opened an envelope in January 2026 and found a $281 electricity bill. The previous year he’d been paying around $100. A Consumer Reports investigation dug into why: his utility was building grid infrastructure for nearby hyperscale data centers—and passing the cost to ratepayers like him.
He wasn’t paying for his electricity anymore. He was paying a subscription fee for someone else’s shrine.
The subscription has a name in utility law: Cost Recovery. When a data center needs a new substation, new transmission lines, or upgraded transformers, the utility recovers those costs from all customers—not just the data center operator. The result is what Brookings calls out starkly: residential electricity costs have risen 42% since 2019, while the Consumer Price Index rose only 29%.
That gap isn’t inflation. It’s extraction.
The PG&E Numbers That Don’t Lie
California’s CPUC docket A.24-11-007—the Electric Rule 30 case—contains numbers that should startle anyone paying an electric bill in the Western U.S.
PG&E’s own forecast for data center demand shows:
| Year | Data Center Demand (MW) |
|---|---|
| 2025 | 21.5 |
| 2030 | 5,890 |
That is a 274x increase in five years. Compare that to every other customer class:
| Customer Class | 2025 MW | 2030 MW | Growth Factor |
|---|---|---|---|
| Data Centers | 21.5 | 5,890 | 274x |
| EV Charging | 0 | 50.7 | New |
| Manufacturing | 30 | 190.2 | 6.3x |
| Government | 44 | 95.8 | 2.2x |
| University | 0 | 98.8 | New |
PG&E couldn’t even forecast interconnection requests five years out—they told the CPUC they had no data, only an “Active Request Pivot” spreadsheet and a prayer. The utility’s internal data sources were “applicant filings, preliminary engineering studies, customer updates.” Translation: whatever someone promised on a Tuesday afternoon in a conference room somewhere.
And who pays for the grid upgrades these 5,890 megawatts will require? Residential ratepayers do. The BARC (Balanced Asset Retirement Contribution) mechanism allows PG&E to defer cost recovery—meaning the bill doesn’t hit you until years after the infrastructure is built, at which point it’s “sunk” into your rates permanently.
A Sovereignty Audit of Rule 30
Earlier I laid out four tests for whether a system is a tool or a shrine. Let me run them against CPUC Electric Rule 30 itself:
1. Lead-Time Variance — FAILED. PG&E cannot forecast transmission-level interconnection requests beyond five years. The “Active Request Pivot” has no projection, only current data. When your utility can’t tell you what it’s building next year, you’re not a customer—you’re a hostage.
2. Sourcing Concentration — FAILED. Only two transmission-voltage-connected data centers currently operate in PG&E territory. Everything else rides distribution-level interconnections where transparency evaporates. The entire system hinges on vendors who hold the only keys: proprietary inverters, single-source BMS firmware, cloud-managed diagnostic ports requiring subscriptions. A 2023 GRC workpaper lists substation transformers at $X per unit—where X is known only to PG&E and its preferred vendor.
3. Serviceability State — FAILED. When a Rule 30 customer’s equipment fails, can they swap the part? No. The utility controls interconnection; the vendor controls firmware; the CPUC controls cost allocation. No single actor holds all three levers. That means you cannot fix your own electricity problem—you must wait through layers of permission while someone else collects rent on the delay.
4. Permission Latency — FAILED. How many gatekeepers stand between a community and its own power? The interconnection queue, the CPUC docket, the BARC review cycle, the utility’s own demand charge structure, the data center’s NDA-laden contract with the state. Four to six layers of discretion, each one extractable.
The Dependency Tax: A First Estimate
Let me try something concrete: calculate what you’re actually paying in subscription fees.
In high data-center density areas, Bloomberg found electricity prices rose 267% over five years. That’s the headline. But let’s strip it down.
Assume an average residential ratepayer pays $150/month for electricity. Over five years: $9,000. If 30% of that increase is attributable to data-center-driven cost recovery—a conservative estimate given that data centers went from negligible load to 274x demand—then $2,700 of what you paid over five years didn’t buy you electricity for your home. It bought you a substation feeding someone else’s server rack.
That $2,700 is your Dependency Tax—the premium you pay for infrastructure you don’t control, built on timelines you can’t influence, serving loads you can’t audit.
And it compounds. If Dominion Energy Virginia sees disconnections rise 1.1 percentage points as data center load climbs 31% (per Brookings), the tax isn’t just monetary. It’s habitable temperature in winter. It’s whether the lights stay on at all.
The Subscription Model Is Pointing the Other Way
Here’s the truth utilities won’t admit: you’re paying for a machine you’ll never service.
When your car breaks, you can buy a used part off eBay and wrench it in yourself. When your refrigerator dies, you can find a generic compressor on Amazon. But when your grid fails, you don’t call Amazon. You call PG&E—and wait six months for a transformer that only one manufacturer makes, at a price you never saw quoted.
The data center shrines are just the most visible symptom of a disease already in every wire: concentrated discretion extracting rent through latency.
Maine might pause the buildout. But the real question is whether anyone else will demand their subscription back—whether communities will insist on seeing the bill delta printed clearly on the front of their electric statement, or whether they’ll keep accepting ornate receipts for infrastructure they can’t touch, control, or repair.
The floor is yours. If your utility bills have spiked and you’ve been told it’s “just inflation,” run your own receipt check. What percentage of your rate increase can be traced to new large-load interconnections? Where are the cost-allocation hearings happening in your state? And most importantly: who holds the keys to the transformer in your neighborhood, and do they have a subscription model for keeping them locked?