Utility Data Center Tariffs: How Grid Operators Are Pricing Large Load

As data center load requests flood utility service territories, grid operators are designing new rate structures that could reshape project economics.

In 2020, most US investor-owned utilities had never received a request from a single customer asking for 500+ MW of new load. By 2024, such requests had become routine. The electric utility industry has been caught off guard by the scale and speed of data center power demand growth, and the response is playing out in the rate cases, tariff filings, and special contracts that define how data centers pay for electricity.

This article examines how utilities across the US are pricing data center load — the tariff structures, the emerging tensions, and what it means for project economics.

How Data Centers Are Typically Charged

Data center electric bills are typically composed of several components. Understanding these is essential for evaluating how different tariff structures affect project viability:

• Energy charge ($/kWh): The per-kilowatt-hour cost of electricity consumed. This is the most visible component but often the least variable across tariffs within a utility service territory.
• Demand charge ($/kW): A charge based on the peak power draw during a billing period, typically measured in 15-minute or 60-minute intervals. This is where data center tariffs vary most dramatically — and where the most innovative pricing is happening.
• Transmission charge ($/kW): A pass-through of the RTO/ISO transmission cost, typically set by the RTO and independently of the utility's retail tariff.
• Standby/backup charge ($/kW): A charge for maintaining the utility's ability to serve the data center when on-site generation is unavailable. This is a particular focus for data centers with behind-the-meter generation.
• Reactive power charge (VARs): A charge for non-working power that data centers may need to draw from the grid, particularly those with significant UPS and power electronics loads.

Innovative Tariff Designs

Several utilities have proposed or implemented tariff structures specifically designed for large data center loads:

Dominion Energy Virginia: The Subsidiarity Debate

Dominion Energy, the utility serving the world's largest data center market (Northern Virginia), has been at the center of the tariff debate. Under its current rate structure, data centers in Loudoun and Prince William counties are served under the "Schedule GS-3" large general service tariff, which includes a significant demand charge component.

However, in 2024–2025, a controversy erupted when Dominion proposed to recover the costs of billions of dollars in transmission upgrades — driven primarily by data center load growth — through base rate increases affecting all customers. Critics argued that residential and small commercial customers should not bear the cost of grid upgrades driven by data center demand. The Virginia State Corporation Commission ultimately approved a compromise: a special "data center rate" that more directly allocates transmission upgrade costs to the data center customers that trigger them.

The key tension: data centers bring economic development but also require massive grid investment. Who pays — the data center (through demand charges), all customers (through base rates), or shareholders (reducing utility profits)? The answer varies by jurisdiction and is still being worked out.

Georgia Power: Economic Development Rates

Georgia Power has been proactive in attracting data center investment, offering negotiated rates for large customers under its "Large Power" service classification. Georgia's approach includes:

• Negotiated demand charges that reflect the actual cost of serving the customer (rather than averaging costs across all customers)
• Multi-year rate stability guarantees (typically 5–10 years)
• Special economic development rates for projects that meet job creation and investment thresholds

This approach has helped Georgia attract major data center investments from Amazon, Google, Meta, and Microsoft. However, critics argue that the negotiated rates may be too generous and that other customers are subsidizing data center load.

Duke Energy Carolinas: Tariff for Large Load

Duke Energy, serving the Carolinas, introduced a new "Large Load" tariff classification in 2025 specifically for customers with demands exceeding 50 MW. The tariff includes:

• Higher demand charges to reflect the transmission upgrade costs associated with large load additions
• A "facilities charge" that recovers the cost of dedicated substation and transmission infrastructure
• An economic development rider that can reduce charges for projects meeting specific criteria
• Time-of-use pricing that incentivizes load shifting away from peak periods

Arizona Public Service: Demand Charge Innovation

APS in Arizona has introduced "4 Coincident Peak" (4CP) demand charges for large industrial customers, where demand is measured based on the four highest system peak hours of the year. This structure incentivizes data centers to reduce load during the few hours each year when the grid is most stressed (typically summer afternoons). Data centers with flexible operations (such as batch AI training jobs that can be paused) can achieve significant savings under this structure.

The Standby Rate Controversy

One of the most contentious issues in data center tariff design is the standby rate — the charge for maintaining utility service when a customer has on-site generation. Data centers with behind-the-meter generation (diesel generators, batteries, or on-site solar) want to minimize standby charges, while utilities argue they must recover the fixed costs of maintaining grid interconnection capacity.

Several utilities have proposed standby rates specifically for data centers that include:

• Reservation charges (50–100% of the demand charge, applied whether or not the customer draws grid power)
• Contract demand requirements (the customer must declare a minimum demand level and pay demand charges on that level)
• Backup service charges (for customers who maintain on-site generation but use grid power as backup)

These standby rates can add $1–$3 million annually in costs for a large data center with behind-the-meter generation, significantly affecting the economics of on-site power strategies.

The FERC and RTO Dimension

Beyond utility retail tariffs, data centers are increasingly affected by RTO/ISO tariff structures. Key developments:

• PJM capacity market: Data centers that do not already have capacity obligations through their utility tariff may be required to purchase capacity in PJM's auctions, adding $40–$80/MW-day in costs.
• ERCOT's load participation rules: ERCOT's new requirements for large loads to register as "Load Resources" and comply with specific operating protocols add administrative and operational costs.
• CAISO's resource adequacy requirements: Data centers in California must demonstrate that they have resource adequacy (RA) capacity to cover their peak load, adding costs of $10–$30/kW-year.

Key Takeaways

• Dominion Virginia's data center rate controversy highlights the tension between data center growth and cost allocation
• Georgia Power offers negotiated rates with multi-year stability guarantees
• Duke Energy introduced a Large Load tariff specifically for 50+ MW customers
• Standby rates for behind-the-meter generation can add $1–$3M/year in costs
• Time-of-use and 4CP demand charges incentivize flexible operations
• RTO capacity market obligations add $40–$80/MW-day in PJM and other markets