Issue 003 - Energy - Growth and unit conversion
How many households equal two years of U.S. electricity-demand growth?
The EIA projects that U.S. electricity consumption will rise from 4,195 billion kWh in 2025 to 4,399 billion kWh in 2027, driven partly by AI data centers, cryptocurrency operations, and electrification. Convert that increase into something easier to picture.
The problem
Estimate how many average U.S. households could be powered for one year by the increase in total U.S. electricity consumption from 2025 to 2027.
Is that increase closer to powering a city, a state, or several states' worth of households for a year?
Because Fermi problems target an order of magnitude, I normally use no more than two significant digits and write most calculations in scientific notation; the Fermi reference explains both conventions.
Before checking sources
Matt's first pass
I was given projected consumption of 4,269 billion kWh in 2026 and 4,399 billion kWh in 2027, so I first calculated the one-year increase:
2026 to 2027 growth ~= 4,399 - 4,269 billion kWh
~= 130 billion kWh
~= 1.3 x 10^11 kWhBecause the question covered growth from 2025 through 2027, I assumed a similar stepping growth and estimated the full increase as 2.3 x 10^11 kWh.
I did not know average household electricity consumption, so I worked backward from a guessed monthly bill of $120 and an electricity price of $0.10 per kWh.
monthly use ~= $120 / ($0.10/kWh)
~= 1.2 x 10^3 kWh/month
annual use ~= (1.2 x 10^3 kWh/month) x (12 months)
~= 1.4 x 10^4 kWh/household-yearhousehold equivalents ~= (2.3 x 10^11 kWh) / (1.4 x 10^4 kWh/household-year)
~= 1.6 x 10^7 household-years
~= 16 million household-yearsMy answer was about 16 million households for one year, which I judged to be closer to a small state or several large cities.
Calibration Score
Matt's Calibration Score: 80 / 100
Higher is better: earn points for accurate pegs, sound models, correct math, and a result close to the sourced answer. The image shows percent full of it: 100 minus the Calibration Score.
Pegs: 20/30. The household-electricity estimate was built well from monthly bill and price intuition.
Model: 30/30. Demand growth divided by household annual use is the right model.
Math: 10/10. The arithmetic was clean.
Result: 20/30. The answer was close, with the main household-use peg off by only a modest factor.
Grounding facts
Nineteen million households is unmistakably multi-state scale. Census QuickFacts reports about 8.75 million households in Florida and 7.72 million in New York, or roughly 16.5 million combined. The projected national increase could cover that combined household count for a year, with several million household-years left over.
Matt's cost-based route was sensible, but the assumptions are worth updating: EIA reports a 2024 national average residential bill near $144 per month, an average price near 16.5 cents per kWh, and consumption near 865 kWh per month. Together those imply roughly 10,400 kWh per year.
After checking sources
Check and recalibrate
The July 2026 EIA forecast provides the missing baseline: 4,195 billion kWh in 2025. Subtracting that directly from the 2027 projection gives:
demand growth ~= (4,399 - 4,195) billion kWh
~= 204 billion kWh
~= 2.04 x 10^11 kWhEIA's rounded household anchor is about 10,500 kWh per year, or roughly 1 x 10^4 kWh for Fermi work.
household equivalents ~= (2.04 x 10^11 kWh) / (1.05 x 10^4 kWh/household-year)
~= 1.94 x 10^7 household-years
~= 19 million household-yearsThe sourced result is about 19 million household-years, remarkably close to Matt's 16 million. His demand-growth estimate was about 13% high, while his inferred household use was about 37% above the national average; those errors partly offset.
One conceptual caution matters: this comparison does not mean the additional electricity will literally go to homes. It converts total economy-wide growth into household equivalents to communicate scale.
Post-check reflection
Matt's reflection
My intuitions were pretty close on this one. I am happy that my monthly household usage estimate came out so close to correct, especially since I worked it out backward from a guessed monthly bill and cost per kWh. Being off by 37% is not terrible for this kind of problem.
The corrected answer does not feel impossibly large to me, but it is still a major scale. What I am especially curious about is how much of that increase will come from AI data centers, cryptocurrency operations, and other electrification. I wonder if this deserves more attention so we can understand how our energy allocation is shifting in recent years.
Recommended memory peg
Remember that an average U.S. household uses roughly 1 x 10^4 kWh per year. It is a useful anchor for translating grid-scale energy figures into a familiar unit.
Reader results
Bars show how submitted estimates sort into the answer choices from the gut-check prompt.