← Briefing history

Grid operators and state regulators are shifting the financial risks of the AI buildout directly onto tech developers through aggressive…

Read-only snapshot of The AI Power Bill

Aug 24, 2026 · 2 findings · ran 4m 18s

TL;DR

Grid operators and state regulators are shifting the financial risks of the AI buildout directly onto tech developers through aggressive tariff restructurings and capacity market exclusions. From Oregon's landmark "growth pays for growth" rate increases to PJM's emergency backstop auctions and large-load curtailment rules, the era of socializing data center connection costs is coming to an end.


The Transition to "Growth Pays for Growth" Rate Structures

Regulators are dismantling the traditional socialization of grid expansion costs by establishing separate, high-tariff customer classes for large-scale data centers.

"ADS requests rehearing or reconsideration of the portion of the Order directing Portland... For each of these reasons, ADS respectfully requests reconsideration or rehearing of the LEUFonly interconnection queue portion of the Order." — Amazon Data Services Application for Rehearing or Reconsideration (PDF)

Oregon's implementation of the PGE Schedule 96 tariff under the state's 2025 POWER Act represents a major regulatory pivot, forcing facilities over 20 MW into a separate customer class to protect residential ratepayers [oregon-pwr-act-pge-schedule-96-tariff]. By mandating that massive tech loads fund their own clean energy resources and pay steep premiums, states are forcing hyperscalers to choose between paying higher rates or facing restricted grid access [oregon-pwr-act-pge-schedule-96-tariff].

What to watch: The Oregon Public Utility Commission's ruling on Amazon's application for reconsideration, which is legally due by September 4, 2026 [oregon-pwr-act-pge-schedule-96-tariff].


Grid Operators Insulate Ratepayers via Demand Exclusion and Curtailment

Grid operators are rewriting market rules to force data centers to bring their own power generation or accept severe, system-directed curtailment.

"Existing consumers should not bear higher capacity costs caused by new large loads that do not bring, or otherwise contract for, the new supply necessary to serve them." — PJM Board Directs Action on Resource Adequacy, Affordability and Large Loads

PJM's decision to exclude new large loads lacking dedicated power from future capacity auction forecasts marks a desperate defense against soaring regional capacity costs [pjm-deploys-reliability-backstop-procurement-framework]. Under the newly proposed Interim Resource Adequacy Service (IRAS) framework scheduled for June 1, 2027, data centers must agree to emergency curtailment or run backup generators, transforming them from favored customers into interruptible grid assets [pjm-deploys-reliability-backstop-procurement-framework].

What to watch: Whether FERC approves PJM's proposed backstop capacity auction, which is scheduled to run from September 30 to October 21, 2026 [pjm-deploys-reliability-backstop-procurement-framework].


What surprised us

  • The $29.4 Billion Legacy Cost: PJM's Independent Market Monitor revealed that leaving existing data center loads in past capacity auctions has already cost the region $29.4 billion over just the last four auctions, illustrating why the grid operator is rushing to exclude them from future demand curves [pjm-deploys-reliability-backstop-procurement-framework].
  • The $878/kW Interconnection "Project Killer": Competitive Power Ventures was forced to withdraw its planned Ohio power plant project after receiving an astronomical $878/kW interconnection cost estimate from PJM—a price that historically neared the total cost to construct an entire plant [pjm-deploys-reliability-backstop-procurement-framework].
  • The 90% Take-or-Pay Minimum: Under Portland General Electric's Schedule 96 tariff, data center developers are obligated to pay for at least 90% of their requested capacity even if their actual electricity usage is lower, preventing hyperscalers from "hoarding" grid capacity they do not immediately use [oregon-pwr-act-pge-schedule-96-tariff].

Open threads worth a vote

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.