Data Center Electricity Rates Rise 29%: How Oregon Is Preventing Households from Paying for AI Power Demand
Oregon is reallocating AI infrastructure risk through a dedicated data center tariff, unused-capacity responsibility, and clean-power cost recovery.
Key terms
14 terms used in this guide, kept within a 20-term reading limit.
- POWER Act
- Oregon law assigning large-load grid costs to the customers creating them.
- HB 3546
- The legislative number of the POWER Act.
- Oregon PUC
- The regulator for Oregon's investor-owned utilities.
- PGE
- Portland General Electric, one of Oregon's major utilities.
- Schedule 96
- PGE's dedicated rate class for large data centers.
- Large energy-use facility
- A qualifying data center or other large electrical load.
- Demand
- The highest power requirement during a defined period.
- Peak load
- Electricity use during the system's highest-demand periods.
- Stranded asset
- Grid infrastructure built for demand that does not fully materialize.
- Line extension allowance
- Utility support for infrastructure connecting a new customer.
- Clean electricity
- Emissions-free or low-carbon power meeting state requirements.
- Energy storage
- Batteries or other resources that shift energy across time.
- Load flexibility
- The ability to change electricity demand in response to grid conditions.
- Hourly analysis
- Comparing load, supply, and cost for every hour.
The 29% increase is not simply a penalty
Beginning in July 2026, large data centers and other large users served by PGE saw an average 29% electricity-rate increase. A later state report said average residential rates decreased by about 1.3%. The policy objective is to assign the cost of substations, lines, reserves, and clean power to the customers creating the demand.
- Utilities must build infrastructure before full load arrives
- Delayed or downsized projects can leave underused assets
- Ordinary tariffs can shift those costs to households and small businesses
Research note: Oregon is changing risk allocation, not merely the price per kilowatt-hour.
How the POWER Act reallocates cost
Oregon enacted HB 3546 in 2025. It requires a separate customer class for large energy-use facilities and directs utilities to assign service costs to those customers. PGE then created Schedule 96.
- Large data centers are priced according to their load characteristics
- Predictable charges apply when contracted infrastructure is not fully used
- Customers pay for clean electricity required to serve incremental load
- Facilities allocated 100 MW or more pay an additional one cent per kilowatt-hour
Research note: Surcharge revenue supports programs intended to reduce residential energy burdens.
Four solutions and their limitations
The Oregon case combines capacity commitments, connection costs, load flexibility, and new supply rather than relying on one average tariff.
- Dedicated tariffs and long contracts reduce speculative reservations but increase fixed cost
- Higher upfront connection payments reduce cost shifting but raise project capital requirements
- Efficiency and flexibility can reduce peaks, but real-time inference and cloud services cannot always curtail
- New clean power and storage connect demand with supply, but annual procurement does not ensure hourly reliability
Research note: No single instrument can solve affordability, reliability, and decarbonization at the same time.
A higher tariff does not create grid capacity
Fairer rates protect other customers but do not immediately create substations, transmission, or firm generation. Requested capacity may also exceed eventual demand, while multiple projects in the queue can cause utilities to over-plan the system.
- Connection reviews should test project maturity
- Capacity commitments need milestones and exit responsibility
- Annual energy figures must be supplemented by peak and nighttime analysis
Research note: Tariffs are the governance entry point; hourly load and supply data still drive the decision.
Six questions developers should answer
Electricity pricing is now a site-selection and investment issue rather than a cost discovered after commissioning.
- How much requested capacity will actually be used
- Which computing workloads can avoid system peaks
- Whether incremental demand is matched by incremental generation
- How long storage can support the facility
- What remains payable if the project is delayed or downsized
- Whether total energy, connection, and clean-power costs remain competitive
Research note: Earlier scenario modeling reduces the risk of securing capacity that later proves uneconomic.
The KUNIES perspective From annual claims to hourly decisions
KUNIES helps data center, renewable-energy, and storage stakeholders integrate hourly load forecasts, renewable and battery simulations, peak tariffs, demand charges, and gaps between power contracts and actual consumption.
- Compare total power costs across sites and capacities
- Quantify peak-cost reductions from storage and load shifting
- Track supply source and shortfall during critical hours
- Maintain shared evidence for investment, operations, and audit
Research note: Securing large-scale electricity for AI infrastructure now requires a credible plan for grid capacity and clean power, not only capital and land.
Research sources
These sources support the market context and operating methods in this guide. Project decisions should use current local rules, contracts, and engineering evidence.