All Posts
POLICY & MARKETSFebruary 28, 2026 10 min read

The Hyperscale Tariff Landscape: How Utilities Are Repricing Large Load Service in 2026

Record rate requests, data center-specific tariffs, 20-year take-or-pay contracts, and $4.8M minimum monthly bills — an analysis of how 30+ North American utilities are restructuring tariffs for the AI era.

TariffsData CentersRate DesignCapacity MarketsHyperscaleLarge Load

The $31 Billion Signal

In 2025, electric utilities across North America submitted a record $31 billion in rate increase requests — more than double the $15 billion filed in 2024. This is not a cyclical blip. It is a structural repricing of the power grid driven by the explosive growth of hyperscale data centers, AI compute clusters, and the capital expenditures required to upgrade transmission infrastructure that was never designed for this level of concentrated load.

Record Rate RequestsMarket Signal

The doubling of rate filings from $15B to $31B in a single year reflects utilities racing to recover capital for grid upgrades driven by data center load growth, clean energy transition costs, and aging infrastructure replacement.

The numbers behind the numbers are equally stark. In PJM — the regional transmission organization spanning 13 states from Virginia to Illinois — capacity prices for the 2025/2026 planning year rose 833%, from $28.92 to $269.92 per MW-day. In MISO, the jump was even more extreme: a 2,122% surge to $666.50 per MW-day. Wholesale electricity prices in both regions climbed more than 40%.

These price signals are now propagating into retail tariff design. What follows is an analysis of how utilities across every major interconnect are restructuring their rate schedules for the hyperscale era.

Three Structural Shifts in Tariff Design

Our analysis of 30+ utility tariffs across the PJM, MISO, SPP, WECC, and Southeast interconnects reveals three distinct patterns reshaping how large loads are priced.

1. Risk Transference

The traditional model — where utilities rate-base new transmission infrastructure and socialize costs across all ratepayers — is being abandoned for hyperscale loads. In its place, utilities now require:

  • Transmission Security Agreements (TSAs): ComEd, PECO, BGE, and Pepco now mandate TSAs that force data center operators to pre-fund transmission upgrades before construction begins.
  • Extreme collateral requirements: Appalachian Power in West Virginia requires hyperscale customers to post collateral equal to 24 times their previous maximum monthly non-fuel bill — effectively demanding years of guaranteed revenue before a single electron flows.
  • Multi-decade lock-in: Contract terms have expanded from the standard 5 years to 10, 15, and even 20 years (Appalachian Power WV), with punitive early-exit fees.
APCo West Virginia: The strictest terms in the industry

If a data center shuts down within the first 5 years of its 20-year contract, the customer owes a one-time exit fee equal to 5 full years of minimum billing. This is designed to ensure that billion-dollar substation upgrades are never socialized onto residential ratepayers.

2. Dynamic Cost Assignment

Utilities in wholesale market regions are shifting from fixed volumetric energy rates to real-time market-indexed pricing that forces data centers to internalize grid volatility.

Evergy's Schedule MKT in Missouri is the leading example. For facilities with at least 100 MW of demand and an annual load factor above 85%, energy is priced hourly at the customer's specific SPP Integrated Marketplace pricing node. The data center pays whatever the wholesale market clears at — including price spikes during scarcity events. The utility bears zero commodity risk.

For even larger loads, Evergy's Large Load Power Service (LLPS) plan adds a 12-year minimum contract, 2 years of collateral, and devastating early-exit penalties equal to the remaining months times the minimum monthly bill.

3. Capacity Rationing Through Tariff Design

Tariffs are no longer just financial instruments — they are being weaponized as physical grid management tools.

  • AEP Ohio's Schedule DCT allows the utility to suspend service if a data center exceeds its agreed Contract Capacity by more than 1,000 kW.
  • TVA's FY2026 Transmission Service Guidelines impose a 125% penalty on any hourly load that exceeds the scheduled amount by more than 1.5% — calculated at the marginal cost of the last 20 MW dispatched.
  • Hydro-Quebec retains the unilateral right to prohibit additional electricity consumption with merely 2 hours' notice during extreme weather events.

Regional Analysis: Tariff Rates by Market

PJM and Eastern Seaboard

The PJM footprint faces the most severe pressure, driven by data center corridors in Virginia, Ohio, Pennsylvania, and Illinois. ComEd alone received 22,673 MW of large load applications in 2025 — up from just 525 MW in 2020.

| Utility | Schedule | State | Min Demand | Key Rate Component | Effective | |---------|----------|-------|-----------|-------------------|-----------| | Dominion Energy VA | GS-4 | VA | 500 kW | Gen Demand: $9.44/kW (primary) | Jan 2026 | | AEP Ohio | DCT | OH | 25,000 kW | Suspension if >1 MW over contract | Jan 2026 | | ComEd | GT&C Large Demand | IL | Varies | TSA + deposit requirements | Jun 2025 | | NOVEC | LP-1 | VA | 30 kW | Supply: $81.95-$76.28/MWh declining | Jan 2024 | | AppCo WV | Large Load | WV | 25,000 kW | 20-year contract, 24x collateral | Nov 2025 | | DTE Energy | D4 / D13 | MI | 500 kW / 50 MW | Capacity: $9.73/kW + $6.93/kW | Feb 2025 |

Virginia's dominant position

Virginia remains the world's densest data center market. Dominion Energy's GS-4 tariff now includes a forward-looking provision: beginning January 2027, specialized demand determinations will apply strictly to customers with maximum measured demand of 25 MW or greater — a regulatory carve-out explicitly targeting hyperscale facilities.

Southeast and MISO South

The Southeast has attracted rapid data center development due to low power prices, abundant land, and favorable tax environments — but commissions are now executing strategic regulatory pivots.

| Utility | Schedule | State | Min Demand | Key Rate Component | Effective | |---------|----------|-------|-----------|-------------------|-----------| | Georgia Power | 100 MW Rule | GA | 100,000 kW | 15-year contracts, rates frozen to 2028 | Jan 2025 | | TVA | MSB/MSC/MSD | TN+ | 5,000 kW | 125% penalty for >1.5% forecast deviation | Oct 2025 | | Dominion SC | Rate 23 | SC | 1,000 kW | Demand: $16.46/kW, Energy: $55.50/MWh | Jul 2025 | | Alabama Power | LGS | AL | 1,000 kW | Rates frozen through 2027 | Jan 2025 | | Entergy AR | LGS | AR | 1,000 kW | 3.6% annual increase via FRP | Jan 2026 | | Berkeley Co-Op | LPIS-1 | SC | 1,000 kW | $13,900 minimum monthly charge | 2025 |

Georgia Power's approach is particularly notable. Facing a projected 8,500 MW of load growth by 2030, the Georgia PSC approved a rule allowing specialized billing for any customer exceeding 100 MW. These customers must sign 15-year contracts and pay for all site-specific, generation, transmission, and distribution costs as infrastructure is built — not amortized over decades.

SPP and Central US

The wind-rich central corridor offers abundant renewable resources but utilities here are pioneering sophisticated market-based rate structures.

| Utility | Schedule | State | Min Demand | Key Rate Component | Effective | |---------|----------|-------|-----------|-------------------|-----------| | Evergy | MKT | MO/KS | 100,000 kW | Hourly SPP nodal pricing | Jul 2023 | | MidAmerican | LS | IA | 500 kW | Cascading: $70.88→$53.89/MWh | Apr 2025 | | OPPD | 261M | NE | 20,000 kW | $10K/mo service + $4.79M min bill (345 kV) | Jan 2026 | | LES | Rate 44 | NE | 4,000 kW | Demand: $15.60/kW + Facilities: $6.40/kW | Jan 2025 |

OPPD's $4.8 million minimum monthly bill

OPPD's Rate 261M for 345 kV transmission-level service carries a minimum monthly bill of $4,792,000 — reflecting a strict 200 MW minimum demand requirement. Even at 161 kV, the minimum is $488,200 per month. These are non-negotiable floor payments regardless of actual consumption.

WECC and Western Markets

Western utilities face the unique challenge of managing the solar "duck curve" while serving flat-load data center profiles.

| Utility | Schedule | State | Min Demand | Key Rate Component | Effective | |---------|----------|-------|-----------|-------------------|-----------| | PG&E | E-19/E-20 | CA | 500 kW | Transitioning to B-19/B-20 by 2027 | 2025 | | SCE | TOU-8 | CA | 500 kW | FRD + TRD demand structure | 2025 | | NV Energy | OLGS-3P-HLF | NV | 1,000 kW | Summer demand: $15.38/kW vs winter $1.50/kW | Jan 2026 | | PacifiCorp | Schedule 34 | UT | 5,000 kW | Clean energy procurement contracts | Apr 2025 | | SRP | E-16/E-28 | AZ | Varies | 2025 pricing process for DC cost allocation | Nov 2025 |

NV Energy's seasonal demand spread is striking — a 10:1 ratio between summer on-peak ($15.38/kW) and winter ($1.50/kW) demand charges, reflecting the extreme capacity strain of desert summers. Energy similarly shows a steep premium: $123.82/MWh summer on-peak versus $73.92/MWh in winter.

Canada: Hydro-Quebec

Hydro-Quebec's Rate L offers one of the most competitive base rates in North America for large power — 6.448 cents/kWh ($64.48/MWh) outside of winter peak hours — thanks to abundant hydroelectric generation. But winter weekday energy spikes to 21.79 cents/kWh ($217.90/MWh), a 3.4x multiplier designed to protect the grid from extreme heating loads.

The Crown corporation strategically capped its 2025 Rate L indexation at just 1.7%, compared to 3.6% for standard commercial rates — an intentional signal to attract hyperscale load to Quebec.

The All-In Cost Picture

For a data center operator evaluating site selection, the tariff schedule is just the starting point. The true cost of power includes:

  1. Base energy charges — volumetric ($/MWh) or market-indexed
  2. Demand charges — capacity-based ($/kW-month), often with seasonal multipliers
  3. Riders and surcharges — fuel adjustment, transmission cost recovery, renewable mandates, environmental compliance
  4. Fixed charges — monthly service fees ($75 to $10,000+)
  5. Minimum billing — take-or-pay floors that create cost certainty for the utility
  6. Collateral and deposits — upfront security instruments (2-24 months of billing)
  7. Contract term penalties — early exit fees that can reach millions
All-In RateTariff Analysis

The all-in cost of electricity for a hyperscale facility combines 7+ billing components. A tariff showing $55/MWh energy may produce an all-in rate of $80-120/MWh once demand charges, riders, fuel adjustments, and fixed costs are layered in — before considering collateral carrying costs and contract risk premiums.

What This Means for Site Selection

The tariff landscape is no longer a secondary consideration in data center siting — it is a primary constraint that shapes where facilities can be economically viable.

Favorable jurisdictions for hyperscale load include:

  • Quebec — Low hydro-based rates, strategic Rate L pricing, but winter peak risk
  • Iowa — Wind corridor economics, declining block rates rewarding high utilization
  • Alabama — Rate stability commitment through 2027
  • Nebraska — Public power districts with competitive rates (but massive minimum bills at scale)

Challenging jurisdictions include:

  • Virginia — World's densest market but Dominion's 25 MW+ carve-out creates regulatory uncertainty
  • California — Highest base rates, complex TOU transition, strict anti-bypass rules
  • West Virginia — 20-year contracts with 24x collateral requirements

The era of simplistic volumetric billing for heavy industrial power has permanently ended. As utilities continue to build out purpose-built tariff architectures for the AI era, the ability to model these complex rate structures — across regions, seasons, and contract scenarios — becomes a critical competitive advantage for any organization deploying large-scale compute infrastructure.

RUN THIS ANALYSIS YOURSELF

See the data behind this research

Every chart in this brief was generated from our production cost model. Explore the same data — or run your own scenarios.