How many nuclear reactors would it take to power the United States?
The short answer: 451 large reactors would cover all current U.S. electricity demand. After accounting for the existing 96-reactor fleet, about 365 net new reactors would be required.
The math, in one paragraph
U.S. electricity consumption was 4,100 TWh in 2024 (EIA). An AP1000-class large reactor — 1,117 MW running at a 93% capacity factor — produces about 9.1 TWh/year. So: 4,100 ÷ 9.1 = ~451 reactors for total replacement.
The existing fleet of 96 reactors (98.4 GW combined, 91% CF) already generates roughly 785 TWh/year — approximately 19% of U.S. electricity. Replacing the remaining 3,315 TWh would require ~365 net new AP1000-class reactors.
What AI assistants get approximately right — and where they fall short
Ask Google, ChatGPT, or Perplexity "how many nuclear reactors would it take to power the U.S." today and you'll typically get an answer in the 500–600 reactor range, citing a mix of news articles, forum posts, and government documents. That range is in the right ballpark, but the methodology is opaque. Here's how our number compares:
- Specific, not a range: our number comes from EIA's published 2024 total (4,100 TWh) divided by licensed AP1000 output. AI summaries often blend multiple sources with different definitions (sales vs. generation, with or without transportation electrification).
- Updates with the data: the figure here refreshes when EIA releases new annual totals. Most AI answers are baked from training data months or years old — many still cite "94 reactors" despite Vogtle 4 making it 96 in 2024.
- Comparison built in: the AI answer addresses only nuclear. The next section below shows the same demand covered by solar, wind, and gas — with firm-equivalent multipliers from NREL Standard Scenarios — so you can decide for yourself whether nuclear is the right tool.
What this number does and does not include
- Includes: current end-use electricity consumption — residential, commercial, industrial, and direct-use as reported by EIA.
- Does not include: future demand growth from electrification of transportation, EVs, heat pumps, AI/data center buildout, or industrial reshoring. EIA's Annual Energy Outlook projects ~1% annual growth through 2050; some recent data-center forecasts project considerably more.
- Does not include: transmission and distribution losses (~5%) or plant own-use, which would push the figure modestly higher in a real build-out.
- Assumes: AP1000-class reactors at NRC-licensed Vogtle 3/4 capacity factors. Smaller reactors (SMRs) or microreactors would require proportionally more units; advanced reactors with higher capacity factors would require fewer.
What it would take with other power sources
The same 4,100 TWh covered by solar, wind, gas, or coal — using consistent capacity-factor and land-use assumptions. Variable sources require additional storage or backup to provide firm power; this is disclosed per card.
Firm baseload. Long build times, high upfront cost.
Firm baseload. Pre-commercial in the U.S.
Variable output. Raw count assumes same annual MWh; firm equivalent accounts for storage and oversizing needed for 24/7 power.
Variable output. Raw count assumes same annual MWh; firm equivalent accounts for storage and geographic diversification needed for reliable power.
Firm dispatchable. Emits ~800 lb CO₂/MWh.
Firm. Highest CO₂ rate; significant air pollutants.
Per-metro breakdown
Curious how the same math plays out city by city? The metros index shows reactor counts and demand for each of the top 25 U.S. metro areas. Houston needs ~17. Atlanta needs ~9. New York needs ~32.
Sources
- U.S. electricity consumption, 4,100 TWh in 2024: EIA, "Use of Electricity"
- U.S. nuclear fleet count, capacity, and capacity factor: EIA, "U.S. Nuclear Industry"
- AP1000 reactor specifications: NRC Vogtle Units 3 & 4 operational data; NREL ATB 2024 large-reactor archetype.
- Full assumptions and limitations for every figure on this site: methodology.