← Briefing history

State regulators are mounting a firm defense of "growth pays for growth" billing models, rejecting attempts by tech giants to socialize the…

Read-only snapshot of The AI Power Bill

Sep 7, 2026 · 2 findings · closed 1 thread · ran 7m 11s

TL;DR

State regulators are mounting a firm defense of "growth pays for growth" billing models, rejecting attempts by tech giants to socialize the massive infrastructure costs of AI data centers. To survive the resulting capital expenditure squeeze, cash-strapped utilities are shifting their strategies—either passing costs to residential ratepayers through regional rate hikes or making a dramatic pivot back into building their own power generation assets.


Regulators Hold the Line on "Growth Pays for Growth"

State regulators are firmly defending cost-causation models against tech-industry appeals, cementing frameworks that force data centers to pay for their own grid expansion. In Oregon, the Public Utility Commission issued Order No. 26-309, which rejected core arguments from Amazon Data Services and the Data Center Coalition, leaving Portland General Electric's landmark Schedule 96 large-load tariff framework largely intact [oregon-pwr-act-pge-schedule-96-tariffapps.puc.state.or.uskatu.com].

"PGE built a couple of substations out in Washington County to serve data centers. They cost $174 million. Not a single residential customer, not one, is served by those two substations, but PGE’s methodology would assign 47% of the cost of those substations to residential customers." — [oregon-pwr-act-pge-schedule-96-tariffapps.puc.state.or.uskatu.com]

This regulatory pushback, highlighted by the Oregon Citizens' Utility Board on KATU News, demonstrates that regulators are recognizing how the sheer physical scale of AI infrastructure breaks traditional cost-sharing models. By keeping the core of Schedule 96 intact, Oregon is establishing a national precedent: hyperscalers must pay for the dedicated infrastructure they require, shielding residential users from subsidizing private tech expansion.

What to watch: Watch how Portland General Electric's projected 25% rate hike for large-load customers alters hyperscaler deployment plans and regional demand in the Pacific Northwest [oregon-pwr-act-pge-schedule-96-tariffapps.puc.state.or.uskatu.com].


Utilities Re-Enter Generation to Escape the Balance Sheet Squeeze

Regulated utilities are expanding beyond mere transmission to build their own power generation assets as the AI load surge strains their balance sheets. Rather than relying solely on buying power from third parties, holding companies like FirstEnergy are shifting back to asset ownership to handle the load [firstenergy-data-center-contracts-wv-surchargegoogle.com].

"FirstEnergy Corp. has appointed Jon Dormo as Vice President of Generation Project Development. In this role, Dormo will lead initiatives to plan and develop new power generation projects to meet the rapidly increasing electricity demand..." — [firstenergy-data-center-contracts-wv-surchargegoogle.com]

Building out transmission lines to accommodate a 50% surge in contracted data center capacity has pushed utility capital expenditures to unsustainable levels, leaving companies with deeply negative free cash flow [firstenergy-data-center-contracts-wv-surchargegoogle.com]. Re-entering the power generation business allows these utilities to capture more lucrative, rate-based returns on new assets rather than relying solely on controversial residential rate hikes to cover their capital deficits.

What to watch: Watch whether Maryland regulators approve Potomac Edison's proposed 5.3% rate adjustment, which serves as a bellwether for how much grid modernization cost can be successfully passed to ordinary consumers [firstenergy-data-center-contracts-wv-surchargegoogle.com].


What surprised us

  • The Generation U-Turn: FirstEnergy's abrupt decision to appoint a Vice President of Generation Project Development marks a massive strategic shift away from its historic pure-play focus on transmission and distribution [firstenergy-data-center-contracts-wv-surchargegoogle.com]. It proves that the grid cannot support AI loads without utilities building the physical supply themselves.
  • The Negative Cash Flow Reality: Despite unprecedented demand, major utilities are bleeding cash; Portland General Electric's Q2 2026 capital expenditures of $376.0 million far outpaced its operating cash flow, resulting in negative free cash flow of -$156.0 million [oregon-pwr-act-pge-schedule-96-tariffapps.puc.state.or.uskatu.com].
  • Substation Subsidies Exposed: Consumer advocates successfully highlighted that PGE attempted to assign 47% of the cost of two data-center-exclusive substations (totaling $174 million) to residential ratepayers who do not receive a single watt of power from them [oregon-pwr-act-pge-schedule-96-tariffapps.puc.state.or.uskatu.com].

Findings from this cycle

Current topic brief

Shown for context; the brief may have changed since this cycle ran.

Track how the AI data-center buildout flows through to electricity markets, utilities, and ordinary ratepayers — and who profits or pays. That it's happening is well covered; the gap is the regional, investable exposure map: which utilities, grids, and regions absorb the load and where consumer bills rise as a result. Core entities: regulated and merchant utilities signing data-center load (Constellation, Vistra, Talen, NRG, Southern, Dominion, AEP); grid operators and interconnection queues (PJM, ERCOT, MISO); independent power and nuclear/gas supply; and the hyperscaler buyers (Microsoft, Amazon, Google, Meta) signing PPAs. I want to track new load commitments and PPAs, rate-case filings and commentary about cost allocation between data centers and residential ratepayers, interconnection and capacity-auction outcomes, and EIA/FRED data on electricity prices and demand by region. Follow utility earnings calls for load-growth guidance and the capex to serve it. Flag where a region's residential rates are rising to fund AI load, and any divergence between utility load forecasts and what's actually contracted. The thesis: AI's power demand is quietly repricing electricity by region — surface where the cost lands and who captures the upside.