How much electricity does a data center use?

A data center's draw scales with its size. A 100 MW campus — a typical hyperscaler facility — running around the clock at a power usage effectiveness of about 1.12 uses on the order of 1 million MWh per year, roughly the annual electricity of about 90,000 homes. Use the calculator above to estimate any size, in your state.

How much water does a data center use?

Evaporative-cooled data centers consume water twice: directly for cooling and indirectly through the power plants that supply them. This model estimates roughly 2 gallons per kWh of facility energy, so a 100 MW campus is on the order of 2 billion gallons a year. Actual use depends on cooling type and local climate.

How much should my community negotiate in a community benefit agreement?

As a starting benchmark, this model targets 2% of estimated project investment — about $2 million per MW — as an annual community-benefit floor. For a 100 MW project that is roughly $4 million a year. Treat it as a negotiating target, not a cap.

Community calculator

What does a data center cost your community?

Pick a size and your state. The calculator estimates annual electricity, water draw, carbon emissions, rate pressure, and the CBA target your community should be negotiating — all sourced from the same model the full toolkit uses.

Facility size

100 MW

10 MW = small edge facility · 100 MW = typical hyperscaler campus · 500 MW = mega-project

Your state

—annual electricity (MWh)

—homes equivalent

—annual water (million gal)

—annual CO₂ (metric tons)

Economic impact

Rate pressure

Residents pay —× what the data center pays per kWh (— vs $0.05).

CBA target

At 2% of estimated investment, your community should negotiate at least —/year in community benefits.

Grid carbon: — gCO₂/kWh · Water stress: — · Cooling: evaporative (PUE 1.12)

Water draw is scaled by state water stress (×0.85 low, ×1.0 medium, ×1.4 high) — hot/arid sites lose more water per kWh to evaporation than cool/humid ones (Lei & Masanet — Climate- and technology-specific PUE and WUE estimations for U.S. data centers using a hybrid statistical and thermodynamics-based approach (Resources, Conservation and Recycling 182, 2022); state stress ratings: World Resources Institute — Aqueduct 4.0 Water Risk Atlas (baseline water stress by basin/state)).

What to do with these numbers

Print this page, or use the full toolkit to generate a complete action pack — a meeting brief, comment scripts, letters to officials, and a community flyer with these numbers baked in.

Generate your action pack → The health risks, sourced

As of May 2026 (3m ago) · changes weekly — re-check before citing · Source: EIA — Electric Power Monthly, Table 5.6.A: average retail price of electricity to residential customers by state (the current figure, updated monthly)

**Rates** are EIA Electric Power Monthly Table 5.6.A, May 2026 — all 51 refreshed together from the published table rather than spot-edited, so cross-state comparisons stay valid. Note this is a single month, not an annual average: residential prices are seasonal, and May is a shoulder month, so summer-peaking states may run higher than shown. The refresh moved rates up a mean of 18.6%, and 48 of 51 jurisdictions (50 states plus D.C.) rose. The largest increases land in the PJM footprint and other data-center-dense markets — D.C. +61%, Illinois +55%, Maryland +36%, Ohio +36%, New York +34% — which is the correlation worth raising at a hearing, though correlation is all it is: this table cannot attribute a rate increase to a cause. **Carbon intensities** are state annual averages and move far more slowly than prices; they were not part of this refresh and remain broadly usable. **Water stress** is a coarse low/medium/high banding, not a basin-level assessment — a state marked 'low' can still contain a stressed watershed, which is the scale that actually decides a cooling permit.