Power Analysis for Data Center Sites: What to Model, Where to Get the Data, and How AI Helps
Power is the deciding variable in data center site selection — here is how to analyze it systematically before committing to a site.
A site can check every other box — size, zoning, fiber, water — and still die on power. The grid analysis determines whether a project is feasible, what it will cost, and how long it will take to deliver. Most developers who lose time on site selection lose it because they do not run a rigorous power analysis early enough.
The Five Components of Power Analysis
Power analysis for a data center site is not a single number. It is a layered assessment of five distinct variables.
1. Available Utility Capacity at the Site
How much load the local utility can serve from existing infrastructure at or near the parcel. This requires a preliminary load inquiry (PLI) with the serving utility, or a desktop analysis of substation loading reports where publicly available. Many utilities publish distribution substation capacity maps; many do not.
2. Transmission and Distribution Constraints
Whether the transmission system feeding the substation has headroom for large commercial load. This requires review of the utility's integrated resource plan (IRP), transmission planning studies, and FERC Form 714 data for annual load reporting. Utilities operating under FERC jurisdiction publish this; municipal utilities and co-ops are less transparent.
3. Interconnection Queue Position and Timeline
If a new substation or transmission upgrade is required, where does the project sit in the interconnection queue? Queue times have extended dramatically since 2022. In PJM territory (covering Virginia, Pennsylvania, New Jersey and the mid-Atlantic), average processing time for large load interconnection studies now reaches 48 to 60 months. ERCOT (Texas) is faster but has its own capacity constraints.
4. Rate Structure and Power Cost
Industrial rate schedules, demand charges, time-of-use pricing and the availability of large-load tariffs directly affect project economics. A site with available capacity but unfavorable rate structures may underperform a site with constrained capacity and a negotiated special contract rate.
5. PPA Viability
For developers or tenants requiring renewable energy commitments, contracting new generation through a power purchase agreement is a function of the interconnection queue for generation resources, not just load. In heavily congested markets, pairing a data center load with a new solar or wind project can add 18 to 36 months to the development timeline.
Where to Get the Data
The primary public sources for power analysis:
FERC EQIS (Interconnection Queue): Full queue data for each RTO/ISO. Filters by capacity, fuel type, status, county. Updated monthly.
EIA Form 860: Utility generator inventory, substation locations, transformer ratings. Published annually.
EIA-861 (Annual Electric Power Industry Report): Utility service territory maps, peak load data, rate schedule filings.
Utility IRPs: Integrated resource plans are filed with state public utility commissions. Virginia, Texas, Colorado and California maintain online IRP document repositories showing planned infrastructure additions.
NERC LTRA (Long-Term Reliability Assessment): Reserve margin projections by region, published annually. Shows which regions are heading toward capacity shortfalls.
State PSC filings: For utilities under state regulation, transmission upgrade applications and load study results are public record.
Most of this data is public. None of it is integrated. Cross-referencing a specific parcel against all of these sources manually takes weeks.
Where AI Helps
AI is most useful in the data aggregation and cross-referencing layer:
Parsing utility IRP documents (often hundreds of pages) to extract planned substation upgrades near a target geography
Cross-referencing EIA-860 substation locations against target parcels and flagging proximity to high-capacity transmission lines
Monitoring FERC queue updates for interconnection studies affecting specific counties or utilities
Extracting large-load tariff schedules from utility rate filings and comparing cost structures across competing sites
Building scenario models that project total power cost under different rate structures, demand profiles and PPA assumptions
AI does not replace a utility pre-application meeting or a power engineer's review of a substation loading study. The human judgment layer involves interpreting what utility capacity figures mean in context, reading between the lines of IRP language, and negotiating the terms of a load service agreement.
The firms getting to site recommendations faster are not doing more analysis. They are running the same analysis with AI-assisted data pipelines that compress weeks of desk research into days.
The Practical Go/No-Go Threshold
At preliminary screening, most institutional developers apply a simple test before ordering a formal utility study:
Is there a transmission substation within 3 miles with rated capacity above 200 MVA?
Does the utility IRP project no major transmission constraints in this service territory through 2030?
Is the utility's average interconnection study time under 24 months for large commercial load?
If any of these fail, the project either needs a significantly longer timeline in the underwriting or a different site. Most site screening processes eliminate 60 to 80% of candidate parcels on power alone, before any other due diligence begins.
That screening pass should happen first. It rarely does, which is why development teams invest weeks on a site only to discover a grid constraint that was visible in public data from day one.
What This Means for Market Selection
Power analysis at the site level is inseparable from power analysis at the market level. The same frameworks apply: which utilities have reserve margin, which IRPs show planned capacity additions, which markets have favorable large-load rate structures.
Markets where the utility has a functioning large-load development program — CPS Energy in San Antonio, AEP Ohio in Columbus, TVA in Tennessee and Alabama — allow developers to move from preliminary inquiry to load service agreement in 12 to 18 months. That timeline advantage is worth more than land cost savings or tax incentives in most underwriting scenarios.
Power analysis is not a diligence step. It is the first filter.