Nuclear Energy Statistics (2026)

The United States runs the largest commercial nuclear fleet on Earth: 94 reactors at 54 power plants in 28 states, each plant with one to four units, totaling 98,441 MW of net summer capacity (see the table below). That fleet generated 784.8 terawatt-hours of electricity in 2025, up from 781.9 TWh in 2024 and near the all-time high (EIA).

Despite that scale, the fleet has shrunk in count. Twelve reactors have permanently closed since 2013, mostly older single units squeezed by cheap natural gas. Capacity held up because operators ran the remaining reactors harder and added the two new Vogtle units in Georgia.

Nuclear supplied 17.7% of US utility-scale electricity in 2025, down from about 18% in 2024 by the same EIA measure, because total generation grew 2.8% while nuclear output rose only slightly (EIA). That makes it the single largest source of firm, always-available low-carbon power in the country, ahead of hydro and comparable to wind and solar combined.

The share has drifted down slightly from a peak above 20% in the late 1990s, not because output fell but because gas, wind, and solar grew faster. In absolute terms, US nuclear generation is near its record, thanks to high utilization rather than new plants.

Nuclear's superpower is uptime. US reactors ran at a 91.0% capacity factor in 2025 (90.8% in 2024), meaning they produced near their maximum output more than nine-tenths of the year, the highest of any source (see the chart below). Geothermal was next at 65.9%, then gas combined-cycle at 58.4% (EIA).

By comparison, coal averaged 48.7%, wind 34.2%, and solar PV 24.4% in 2025, because sun and wind are intermittent and fossil plants cycle with demand. The US fleet has topped a 90% capacity factor every year since 2012, helped by shorter refueling outages and better operations.

Nuclear punches above its weight on carbon. In 2023 it provided 19% of US electricity but about 47% of the country's carbon-free power, more than utility-scale wind and solar combined, because it runs around the clock rather than only when the weather cooperates (NEI).

That output avoids more than 447 million metric tons of CO2 a year that fossil plants would otherwise emit. It is why many decarbonization studies treat keeping existing reactors open as one of the cheapest ways to cut emissions, and why several states now pay to preserve at-risk plants.

US plants range widely in size. Since its new AP1000 units came online, Georgia's Vogtle is the largest at about 4.5 GW across four reactors, passing Arizona's Palo Verde at 3.9 GW, long the biggest US power plant of any kind (see the table below). California's Diablo Canyon runs 2.2 GW.

At the small end, New York's single-unit R.E. Ginna is about 0.6 GW. A typical modern US reactor is around 1 GW, enough to power roughly 800,000 homes, which is why even one unit closing removes a large, steady block of low-carbon supply from the grid.

America's reactors are old. Roughly half of the operating units are more than 40 years past their original startup, and most were licensed for 40 years, so keeping them running requires renewals from the Nuclear Regulatory Commission. Standard renewals already stretch licenses to 60 years.

The frontier now is 80-year operation via subsequent license renewals. Six units have already been approved to run for 80 years, and about 14 more are under NRC review. Life extension is cheap capacity: it defers the need to build expensive new plants for decades.

Vogtle Units 3 and 4 are the first newly built US reactors since 2016 and the only US deployments of the Generation III+ AP1000 design. Unit 3 entered commercial operation in July 2023 and Unit 4 in March 2024, bringing the fleet back to 94 reactors.

The project is also a cautionary tale on cost. Originally budgeted near $14 billion with a 2016-2017 start, it ran years late and its total cost climbed above $30 billion, closer to $35 billion by some estimates. Those overruns are why much of the industry's hope has shifted to smaller, factory-built reactors.

Worldwide, 443 reactors are operable (the World Nuclear Association count, October 2026) or 413 in operation at end-2025 (the IAEA's stricter tally), together providing roughly 377 to 406 GW of capacity across 31 countries (see the table below). The gap between the two counts is mostly Japanese reactors that are licensed but idle.

The pipeline is substantial: 79 reactors are under construction (about 87 GW), another 122 are planned, and 336 are proposed. China alone has 38 under construction, so it dominates new construction, and much of the growth is concentrated in Asia and the Middle East rather than the West.

Nuclear supplied 8.4% of the world's electricity in 2025, down from 8.7% in 2024, because global generation grew 2.7% while nuclear grew 1.1% (IAEA). The national picture varies enormously: France leads at 68.1%, followed by Slovakia at 64.4% and Ukraine at 52.0%, while the US sits at 17.7% and fast-growing China at 5.0% despite its huge build-out (see the table below).

Together with renewables, low-carbon sources crossed 40% of global generation for the first time in 2024. Nuclear's slice has slipped from its 1990s high as electricity demand and renewables grew faster, but in absolute terms nuclear output is now setting records.

The United States is the runaway leader, generating 785.6 TWh in 2025, roughly 29% of the 2,702 TWh of nuclear electricity worldwide (see the chart below). China is second at 449.9 TWh, followed by France at 375.7 TWh, Russia at 204.9 TWh, and South Korea at 175.5 TWh (World Nuclear Association).

China has the deepest pipeline behind it, with 38 of the world's 79 reactors under construction (about 43 GW). The WNA's US figure of 785.6 TWh sits close to the EIA's net figure of 784.8 TWh for 2025.

The construction pipeline is the clearest signal of where nuclear is headed. Some 79 reactors, about 87 GW, are being built right now, overwhelmingly in China, India, Russia's export projects, and the Middle East. The UAE's four-unit Barakah plant, finished in the 2020s, already makes it a top-20 generator.

Western new-build has been slower and pricier, as Vogtle showed. That is pushing utilities and governments toward standardized designs and small modular reactors, betting that repeatable factory production can finally bend the cost curve that large bespoke plants never could.

For the sixth year running, the IAEA raised its long-term outlook, and its September 2026 edition extends the horizon to 2060. From 377.1 GW of operational capacity at end-2025, its low case sees 641 GW by 2050 and 696 GW by 2060 (nearly double), and its high case 1,045 GW by 2050 (up from 992 GW in last year's edition) and 1,284 GW by 2060, about 3.4 times the 2025 level (see the chart and table below). Two in three reactors have run for more than 30 years, so both cases require heavy life extension plus new build (IAEA).

Small modular reactors carry much of the upside: they account for 28% of new capacity added by 2060 in the high case and 23% in the low case, up from 24% and 5% in the prior edition, and about 60% of new capacity in North America. In 2025, seven reactors (2.8 GW) retired while three new ones (3.0 GW) connected to the grid and construction began on 12 (14.3 GW).

SMRs are the industry's big bet: factory-built units of roughly 50-300 MW that promise faster, cheaper, more standardized deployment than gigawatt-scale plants. Market-research estimates put the global SMR market between about $6 billion and $7.5 billion in 2024-2025, growing to anywhere from $8 billion to $16 billion by the mid-2030s.

Those forecasts vary widely and come from private market-research firms (secondary sources), so treat the exact dollar figures as directional. The common thread is fast growth, roughly 5% to 15% annual, driven by clean-energy targets, energy security, and, increasingly, data-center power demand.

The sharpest new driver is artificial intelligence. Data-center electricity use is projected to jump from about 460 TWh in 2024 toward 1,000-1,300 TWh by the early-to-mid 2030s, and operators want firm, 24/7 carbon-free power that nuclear can uniquely supply. Over the past year, Big Tech signed contracts for more than 10 GW of new US nuclear (see the table below).

The marquee deals include Microsoft's roughly $16 billion agreement to restart Three Mile Island (835 MW) by 2028, Amazon's commitment to up to 5 GW of SMRs by 2039, and Google's 500 MW SMR fleet deal with Kairos Power. This corporate demand, more than policy alone, is behind nuclear's sudden financial revival.