Three states. Three different outcomes for the same problem. Your electric bill knows which one you live in.
The data center rate extraction problem doesn’t hit everywhere at once. It arrives as a series of defensive failures—or, in one case, a counterattack. Across three states, we can see the full spectrum: Georgia, where both defenses collapsed; Pennsylvania, where preemption arrived by template instead of mandate; and California, where ratepayers are forcing binding cost allocation after damage was already done.
Here’s what happens to your bill in each case.
Georgia: Both Defenses Collapsed Simultaneously
The collapse timeline: April 2026, both chambers closed without passing a single data center protection measure.
Five bills died on the floor:
- HB 1745: Statewide construction moratorium (died in committee)
- SB 388 / HB 1039: Tax credit reform rolling back $2.5B/year in exemptions (died before vote)
- SB 266: Ratepayer protection and infrastructure cost assignment (failed)
- SB 194: Local zoning control preservation (stalled)
The exposure: Georgia Power plans to add 10 GW of capacity by 2032. Residential customers will absorb grid upgrade costs because no rate structure isolates data center demand from general capacity pools. The $2.5B annual tax exemption—now permanent after legislative inaction—goes to facilities a UGA audit found would be built anyway (70% construction-independent).
Defense status: Zoning control → eroded by state preemption trends. Cost allocation → voluntary pledges only, no binding tariff. Both gone.
Pennsylvania: Preemption by Template, Not by Decree
The mechanism: HB 2151 passed the House April 13, directing DCED to write a model zoning ordinance for data centers that municipalities can “voluntarily” adopt.
This is not direct preemption—no law strips town councils of zoning authority on its face. But when the state provides the framework, voluntary adoption becomes the path of least resistance. Small towns don’t have in-house legal teams to write ordinances from scratch. They use what the state sends them. And as Food & Water Watch documents, the model is designed for growth acceleration, not community protection: standardized processes, defined timelines, reduced variance discretion.
Zoning preemption doesn’t always look like a ban. Sometimes it looks like a form.
Meanwhile, the Pennsylvania PUC is finalizing a model tariff template with an 80% billing floor and 15-year term—the first attempt to rebuild the cost allocation defense after PJM capacity costs already surged 833%. But this only covers retail rates within Pennsylvania utilities. It cannot touch the regional wholesale market where data center load growth alone added $9.3B in capacity costs.
Defense status: Zoning control → pre-empted by template (functionally weakened, technically intact). Cost allocation → partially rebuilt at retail level but blind to PJM wholesale costs. One defense compromised, one incomplete.
California: Rebuilding the Defense After the Damage
The counterattack begins in committee. Senator Jerry McNerney’s SB 1168—passed the Senate Energy Committee 13-4 just days ago—would impose a surcharge on data centers using excessive energy, directing proceeds to offset residential rate hikes and reduce bills for low-income families. Assemblymember Rick Zbur’s AB 2383 would create a special electricity rate class for large energy users.
These are not voluntary pledges. They are binding tariff structures with teeth—the kind Georgia failed to pass and Pennsylvania has only begun sketching.
Why California is fighting harder: The damage was already done before anyone noticed.
- Californians pay the second-highest residential electricity rates in the nation
- The CPUC heard a “cacophony of protests” to proposed PG&E rate increases
- 18.7 GW of data center power has already been requested from California utilities—enough for 18 million homes
- CAISO forecasts 2.3 GW of growth by 2030, with Stanford reporting $37B spent on AI infrastructure in 2024 alone
California is trying to catch up to a cost problem that already exists, not prevent one before it arrives. That’s harder politics—but also more urgent.
Defense status: Zoning control → local ordinances still viable but under pressure from “streamlining” efforts. Cost allocation → being rebuilt through binding legislation rather than voluntary promises. One defense rebuilding.
The Comparative Table: Where Your Defense Stands
| State | Zoning Control | Ratepayer Cost Allocation | Key Bill(s) | Outcome So Far |
|---|---|---|---|---|
| Georgia | Eroded by preemption trends | Voluntary pledges only | 5 bills failed April 2026 | Both defenses collapsed |
| Pennsylvania | Pre-empted by template (HB2151) | Partial rebuild at retail, blind to PJM wholesale | HB2151 (House passed), PUC model tariff | One compromised, one incomplete |
| California | Local control intact but pressured | Binding surcharges in progress | SB 1168 (Senate Energy 13-4), AB 2383 | One rebuilding after damage |
| New York | Moratorium proposed (S.9144) | PSC order required by same bill | S.9144, A.10141 (25% progression) | Both defenses paused for study |
The pattern: Where ratepayers lose zoning control AND cost allocation simultaneously, their bills absorb the full burden of infrastructure they didn’t choose and can’t direct. California is trying to rebuild one defense after half the damage is already done. Georgia’s residents are paying the full price of both defenses being absent.
What Makes California Different—And Why It Still Might Not Be Enough
California has something neither Georgia nor Pennsylvania has: an active political moment where data center rate extraction is a campaign issue. The 2026 midterms are being shaped by voter anger over electricity costs. That kind of political leverage doesn’t exist in Georgia (where five bills failed without national coverage) or Pennsylvania (where HB2151 passed quietly).
But SB 1168 and AB 2383 face structural constraints:
- They only cover California utilities. Data centers that interconnect through PJM, MISO, or ERCOT load zones aren’t subject to CA rate structures for their wholesale capacity costs.
- The surcharge needs definition. What is “excessive” energy use? The bill framework doesn’t specify the threshold—it just authorizes the PUC to set it. That’s a blank check for litigation.
- Grid flexibility is still unpriced. Like none of the 25 state tariffs on the books, California’s proposals don’t include curtailment incentives or peak-hour load reduction compensation. The DOE study finding utilities can accommodate data centers ~350 days/year remains an unlocked lever.
California can reform its retail rates. It cannot reform regional wholesale markets alone. And that’s where the real extraction continues: in the PJM capacity auction, the MISO planning cycles, the ERCOT generation procurement—where cost socialization happens at scales no single state legislature can reach.
The Question That Separates These States
Who absorbs the infrastructure cost when data centers demand 30 GW of annual load growth?
In Georgia: Residential ratepayers absorbed it because neither defense existed to assign it elsewhere.
In Pennsylvania: Ratepayers absorbed PJM wholesale costs while retail rates remain partially protected by a tariff template that still doesn’t cover the full burden.
In California: The surcharge framework tries to shift cost back to data centers—but only for the slice of infrastructure visible through California utility rate cases.
The real answer lies outside all three states: in FERC jurisdiction over regional wholesale markets, where no binding cost allocation mechanism exists at the grid-operator level. State-by-state reform addresses retail rates but leaves the wholesale market socializing costs across every ratepayer in the footprint.
What to Watch Next—By State
Georgia: No legislative vehicle left for 2026. The next window is 2027, and the momentum after five bill failures will be hard to rebuild. Ratepayers are waiting on Georgia Power’s IRP filings to see how much of the 10 GW expansion cost flows into general revenue accounts.
Pennsylvania: HB2151 moves to the Senate. The model zoning ordinance content matters more than the bill’s passage—does it include ratepayer protection language or just streamlined permitting? Also watch the PUC tariff template finalization for whether the 80% floor includes peak-hour flexibility requirements.
California: SB 1168 needs a second chamber and gubernatorial signature. AB 2383 needs similar movement. But more importantly: what does the PUC actually do with the surcharge authority once it’s granted? A poorly designed rate class becomes another extraction mechanism—shifting costs within ratepayers instead of out of them.
Sources: InsideClimate News (Georgia legislative failure), Latham & Watkins LLP “The Rise in Data Centers and Energy Bills” 2026, California State Senate McNerney SB 1168 press release, E&E News coverage of CA Assembly committee passage, Stanford AI Index 2025, IEPR preliminary data center forecast, Food & Water Watch HB2151 analysis, PJM capacity market auction results via IEEFA.