Capacity Factor: The Most Important Number in Energy

Take any power plant. Multiply its rated capacity by the 8,760 hours in a year — that's the energy it would produce running flat-out forever. Divide what it actually produced by that ideal, and you have its capacity factor: the single most honest number for comparing electricity sources.

The US nuclear fleet's capacity factor has averaged around 90% or better every year since the early 2000s — the highest of any generation source, and it isn't close. Recent fleet averages run about 92–93%; the mean across plants on this map is in the same range (the site computes it from real generation data — more below). For comparison, typical US figures run roughly: combined-cycle gas plants near 55–60%, coal around 40–50%, onshore wind in the mid-30s, utility solar in the mid-20s. The gaps have different causes — gas and coal plants often could run more but are dispatched less for economic reasons, while wind and solar are capped by how often the wind blows and the sun shines — but the consequence is the same: one megawatt of nuclear capacity delivers two to four times the annual energy of a megawatt of most alternatives.

Why is nuclear's number so high? Economics and physics point the same direction. Fuel is a small share of a nuclear plant's costs, so unlike a gas plant there's no reason to throttle back when prices dip — every additional hour of output is nearly free energy. And the machine itself prefers steady state: constant temperature, constant chemistry, minimal thermal cycling. So plants run at 100% essentially whenever they are able, and the capacity factor effectively measures one thing: how few days the plant was unable to run. Modern US operations have squeezed unplanned downtime to remarkable lows; most of the remaining gap is planned refueling, which is why the outage page is really the other half of this one.

A capacity factor can even exceed 100% for a stretch. Ratings are quoted as net summer capacity — output on the hottest days, when warm cooling water and air make the steam cycle least efficient. In winter, the same reactor pushes out a few percent more than its summer rating, and a plant that ran perfectly through a cool year can land just over 100% of net-summer-based capacity. If you spot a 101% on this map, it's not an error; it's thermodynamics.

How this site computes it: every plant's figure here is derived from real federal data, not quoted from a brochure — twelve months of net generation as reported to the US Energy Information Administration (EIA), divided by the plant's net summer capacity times the hours in those months. A plant that took a refueling outage during the window will show a dip (a 30–40 day outage costs roughly 8–10 points), so a 92% isn't underperformance — it's usually a refueling year, and the plant's history sparkline will show you the outage itself.

One last framing, because it recurs throughout the economics of the AI era: capacity factor is why a "1,000 MW" deal with a nuclear plant and a "1,000 MW" deal with a solar farm are not remotely the same product. The nuclear megawatts show up at 3 a.m. in January. For a data center that must run every hour of the year, that difference is the entire point — and it's why the offtake-deals layer on this map exists.