Methodology & Data Sources

This tool uses industry-standard methodologies and publicly available data to project electricity cost impacts. Below we explain our models, assumptions, and data sources so you can verify and critique our approach.

Model Overview

Our model projects residential electricity bills under four scenarios:

  1. Baseline: Normal cost growth from infrastructure aging and inflation
  2. Firm Load: Data center operates at constant power, adding 100% of capacity to system peak
  3. Flexible Load: Data center reduces load by 25% during peak hours by deferring non-time-sensitive workloads
  4. Flex + Dispatchable: Flexible operation plus onsite generation to further reduce grid draw during peaks

For each scenario, we calculate infrastructure costs, revenue contributions, and allocate net impacts to residential customers based on market-specific regulatory methods.

About the Flexibility Assumptions

The 25% peak reduction capability is based on EPRI's DCFlex initiative, a 2024 field demonstration at a major data center that achieved 25% sustained power reduction during 3-hour peak events. While theoretical flexibility from workload analysis suggests up to 42% is possible, the 25% figure represents a conservative, field-validated baseline. See the Workload Flexibility Model section for details.

Basic Formula:

Monthly Impact = (Infrastructure Costs − DC Revenue Offset) × Residential Allocation / Customers / 12

Key Terms Explained:

Load Factor: Average power draw ÷ nameplate capacity. A 2,000 MW data center at 80% load factor draws 1,600 MW on average.
Peak Coincidence: Fraction of capacity drawing power during system peak hours. 100% means full contribution to peak; 75% means the facility reduces load by 25% during peaks.
Curtailable: The portion of load that can be temporarily reduced during grid stress events by pausing or deferring non-time-sensitive workloads (e.g., AI training, batch processing).

Firm vs Flexible Load Scenarios:

ParameterFirm LoadFlexible Load
Load Factor80%95%
Peak Coincidence100%75%
Curtailable During Peaks0%25%

Why 95% load factor for flexible? By shifting deferrable workloads (AI training, batch jobs) to off-peak hours, data centers can run at higher average utilization while reducing peak contribution. This is more efficient than running at constant 80% regardless of grid conditions.

Grid Capacity Math: Why 33% More?

If a grid has X MW of available capacity for new load:

  • Firm load (100% peak): Grid supports X MW of data center capacity
  • Flexible load (75% peak): Grid supports X ÷ 0.75 = 1.33X MW of data center capacity

Result: 33% more data center capacity can connect to the same grid infrastructure when operating flexibly, because each MW only adds 0.75 MW to the system peak.

Revenue Offset:

  • Demand charges: $9,050/MW-month (based on coincident peak contribution)
  • Energy margin: $4.88/MWh (utility's wholesale spread on energy sales)
  • Higher load factor = more energy sold = more revenue to offset infrastructure costs

Residential Cost Allocation:

The share of net costs allocated to residential customers depends on the utility's market structure. See the "Market Structures & Cost Allocation Framework" section below for detailed allocation factors by market type (regulated, PJM, ERCOT, etc.).

  • Base allocation: Varies by market (30-40% typical)
  • Calculation method: Weighted blend of 40% volumetric (kWh), 40% demand (peak MW), 20% customer count
  • Dynamic adjustment: As data center adds energy and peak, residential shares shift
  • Regulatory lag: Changes phase in over ~5 years through rate case proceedings
  • Market multipliers: ERCOT applies 0.85× (large loads face prices directly); high PJM capacity prices increase allocation

The baseline trajectory includes 2.5% annual inflation and 1.5% annual infrastructure replacement costs.

Questions or Feedback?

If you have questions about the methodology, want to report an error, or have suggestions for improvement, we'd love to hear from you.

Back to Calculator