The bill decoder: every term in the data-center rate debate

Plain English, one real example each

Capacity market · Base Residual Auction · Delivery year · LDA · Energy charge vs. demand charge · Large-load tariff · Minimum take · Rate case · Rate base & return on equity · Rider · Cost allocation · Ratepayer · Interconnection queue · Load growth · Behind-the-meter · Co-location · Power purchase agreement · Curtailment · Transmission vs. distribution · Default service · Formula rates · Hyperscaler · PUE · Moratorium · Integrated resource plan

Capacity market / capacity charge

Payment to power plants for promising to be available at the grid's peak - insurance that the lights stay on, separate from the energy actually consumed. In PJM the price is set in an annual auction and passed through to every customer's bill, usually inside the supply or default-service rate. See current prices.

Base Residual Auction (BRA)

PJM's yearly capacity auction, held about three years before the delivery year it covers. Its clearing price - in dollars per megawatt-day - is the headline number when capacity costs make the news.

Delivery year

The June-to-May year a capacity auction covers. An auction held in 2026 sets prices customers start paying in June 2028 - which is why today's data-center forecasts show up on bills years later.

LDA (Locational Deliverability Area)

A zone within PJM that can clear at its own, higher capacity price when transmission limits make imports hard. That is why northern Illinois or New Jersey can pay more than the RTO-wide floor.

Energy charge vs. demand charge

Energy charges bill the kilowatt-hours you consume over the month; demand charges bill your single highest draw (kW). Homes mostly pay energy charges; data centers and factories also pay demand charges - which is where 'minimum demand' obligations in special tariffs bite.

Large-load tariff

A special rate class for very large customers (thresholds run 20–100 MW and up) with long contracts, minimum payments and exit fees, designed so grid upgrades built for a data center are not stranded on everyone else if it shrinks or never shows up. See the tracker.

Minimum take / minimum demand

The share of contracted service a large customer must pay for whether or not it uses it - 85% of subscribed capacity in AEP Ohio's tariff, 90% in Portland General Electric's. The core ratepayer protection in most large-load tariffs.

Rate case

The formal proceeding where a utility asks its state commission for new rates. It is effectively the trial that sets your bill: utilities file evidence, consumer advocates and attorneys general intervene, and the commission decides - often for less than the ask.

Rate base & return on equity (ROE)

A regulated utility earns a commission-set profit rate (ROE, typically ~9–10%) on the infrastructure it builds (rate base). More grid built for data centers means more rate base - which is why who pays for those upgrades is the whole fight.

Rider / tracker

A separate bill line item that adjusts outside full rate cases, often automatically - like a fuel or transmission cost tracker. Ohio's Basic Transmission Cost Rider (BTCR), which rose about $7.90/month in April 2026 in AEP territory, is one.

Cost allocation / cost shift

The rules deciding which customer class pays for which grid costs. 'Cost shift' is the claim that costs caused by one class (data centers) are landing on another (households) - the central dispute in most current dockets.

Ratepayer

You - anyone who pays a utility bill. Distinct from taxpayers: utility infrastructure is funded through bills, so 'ratepayer protection' means keeping specific costs off them.

Interconnection queue

The waiting line to plug into the grid - for new power plants and, increasingly, giant new loads. ERCOT's large-load queue alone held 233 GW of requests in December 2025, more than 70% from data centers; queued projects often never get built.

Load growth (GW vs GWh)

GW measures instantaneous demand (what the grid must be built for); GWh measures energy used over time (what you are billed for). Data-center forecasts are usually quoted in GW of peak demand - the expensive kind, because it drives new construction.

Behind-the-meter / self-supply

Generation on the customer's side of the utility meter - a data center running its own gas turbines or contracting its own plants. Laws like Utah's SB132 push very large loads toward self-supply so the shared grid does not carry them.

Co-location

Siting a data center directly at a power plant, buying its output before it reaches the shared grid. Contested because that capacity stops serving everyone else, and because the arrangement can dodge shared grid costs.

Power purchase agreement (PPA)

A long-term contract to buy power from a specific generator at a set price. How hyperscalers lock in supply (and its price) for new data centers without owning the plant.

Curtailment / demand response

Reducing a customer's consumption when the grid is strained - voluntarily (paid demand response) or by order. Texas SB6 conditions large-load interconnection on curtailability; New Jersey's 2026 law requires data centers to curtail before households.

Transmission vs. distribution

Transmission is the high-voltage interstate; distribution is the street-level delivery. They are separate bill components with separate cost-allocation fights - most data-center cost disputes so far are about transmission.

Default service / price to compare

In deregulated states (Ohio, Pennsylvania, Illinois…), the supply rate you pay if you never chose a competitive supplier. Capacity-auction increases land here first, at the annual June reset.

Formula rates

Rate-setting that adjusts annually by formula instead of through full rate cases. Arizona's APS has asked for one - supporters say it assigns fast-moving data-center costs promptly; opponents say it weakens scrutiny.

Hyperscaler

The handful of companies operating data centers at giant scale - Amazon, Microsoft, Google, Meta, and AI labs building with them. One hyperscale campus can demand as much power as a small city.

PUE (power usage effectiveness)

A data center's total power divided by the power reaching its computers - 1.0 is perfect. A rough efficiency yardstick; it says nothing about how big the total draw is.

Moratorium

A temporary pause on approving new data centers, at city, county or state level. New York's Executive Order 62 (July 2026) is the first statewide one, pausing state permits for 50 MW+ facilities.

Integrated resource plan (IRP)

A utility's long-range forecast of demand and how it will meet it - where data-center load projections officially enter the record, and where inflated forecasts get challenged.

How data centers reach your bill → · Capacity auctions → · Tariff tracker →