Nuclear Power: Pros and Cons, Costs, Safety and Waste

No energy source splits opinion like nuclear. Supporters cite near-zero emissions, a tiny land footprint and round-the-clock output. Critics cite cost overruns, accident risk and long-lived waste. Both sides have facts to back them, and the right answer differs from country to country. Below is the evidence on climate, safety, waste, cost, speed and grid fit, so you can judge for yourself.

Nuclear in the global mix

Nuclear plants generated about 2,810 TWh in 2025, around 8.8% of world electricity (Ember / Our World in Data). The share has dropped from about 16–17% in 2000. Output didn’t collapse (it has hovered at 2,500–2,800 TWh for two decades); total demand grew while few new reactors were built outside China. About 410–440 reactors operate in some 30 countries (IAEA PRIS), with roughly 60–70 under construction, around half of them in China.

Data: Ember / Our World in Data, latest year. See every country on the country pages and reactor locations on the power plants map.

Pro: very low life-cycle emissions

A life-cycle assessment counts uranium mining, enrichment, construction, operation, decommissioning and waste management.

Source: IPCC AR5 Annex III; UNECE 2021 finds 5.1–6.4 g CO₂e/kWh for nuclear. France’s grid, about 70% nuclear plus hydro and renewables, runs at roughly a tenth of the world-average carbon intensity. See Carbon Intensity of Electricity.

Pro: firm, dense, small footprint

• Capacity factor. Reactors typically run 80–93% of the hours in a year, and the US fleet averages above 90%. Solar manages 10–25%, wind 25–50%.
• Land use. Few sources use less land per unit of energy. UNECE’s assessment gives roughly 0.3–1 m² per MWh/yr over the life cycle, an order of magnitude less than ground-mounted solar.
• Fuel security. Fuel is compact, and plants keep 1–2 years of it on site. Uranium comes from many countries (Kazakhstan, Canada, Australia, Namibia). Enrichment and fuel fabrication are more concentrated, and Russia holds a large share of global enrichment capacity.
• Grid services. Large spinning turbines provide inertia and voltage support.

Con: cost and construction risk

In most Western countries this is the strongest case against new nuclear. Recent European and US projects:

By contrast, Korea’s APR-1400 builds (including Barakah in the UAE), China’s Hualong One and historical French and Japanese programmes came in far cheaper, typically at 5–8 years per unit. Most analysts draw the same lesson: nuclear becomes affordable through serial construction of a standard design, with a retained workforce and stable regulation.

Levelised cost

Construction runs for years before any revenue comes in, so nuclear is extremely sensitive to the cost of capital. Take a plant at $9,000/kW, 90% capacity factor and a 60-year life:

• At 4% real WACC: CRF ≈ 0.044 → capital ≈ $50/MWh; plus O&M and fuel ($25–30/MWh) → **$75–80/MWh**.
• At 9% WACC: CRF ≈ 0.091 → capital ≈ $104/MWh → ~$130–135/MWh.

Lazard’s 2025 analysis puts unsubsidised new US nuclear around $140–220/MWh, versus roughly $40–80/MWh for utility solar and onshore wind (before firming costs). Financing structures such as government loans or regulated asset base models can halve the cost of capital. Try your own numbers in the LCOE calculator.

Existing plants are different. Once built, reactors produce power at roughly $30–45/MWh. Extending licences to 60 or 80 years usually ranks among the cheapest sources of low-carbon electricity on offer.

Con / context: safety and accidents

Public perception rests on three accidents:

• Three Mile Island (1979, USA): partial meltdown; no attributable deaths.
• Chernobyl (1986, USSR): reactor explosion and fire; ~30 acute deaths and an estimated several thousand eventual cancer deaths according to WHO/UNSCEAR estimates (with wide uncertainty); a 2,600 km² exclusion zone.
• Fukushima Daiichi (2011, Japan): meltdowns after the tsunami that killed ~18,000 people; one radiation-attributed death confirmed by Japanese authorities; over 2,000 deaths attributed to the stress and disruption of evacuation.

Per unit of energy, the overall record is very good:

Source: Our World in Data, based on Markandya & Wilkinson (2007) and Sovacool et al. (2016). The figures leave out economic damage, displacement and the psychological weight of a rare but catastrophic risk. Those costs are real, and societies weigh them differently.

Con / context: radioactive waste

A 1 GW reactor produces about 20–30 tonnes of spent fuel per year. High-level waste is intensely radioactive for centuries and stays hazardous for tens of thousands of years, but it is solid, compact and tracked. The options:

• Interim storage in pools, then dry casks. Safe for decades, with a strong record.
• Reprocessing (France, Russia, Japan) recovers uranium and plutonium for MOX fuel, reducing high-level waste volume.
• Deep geological repositories: Finland’s Onkalo at ~430 m depth in granite is the first; Sweden’s Forsmark repository is licensed; France’s Cigéo is in the licensing process.

The hard part is political and social rather than technical: finding communities willing to host a repository.

Con: speed and flexibility

From decision to operation, new nuclear typically takes 10–15 years in Western countries (5–8 in China and Korea). Wind and solar can be planned and built in 1–5 years. Toward 2030 targets, nuclear will contribute mainly through life extensions and restarts. Reactors can load-follow (French ones adjust output daily), but with fixed costs this high, every hour below full output raises the cost per MWh.

Small modular reactors and advanced designs

SMRs (up to ~300 MW) aim to cut cost through factory production, modular assembly and simpler passive safety. Status as of 2026:

• China’s Linglong One (ACP100, 125 MW) at Hainan has been under construction since 2021 and is nearing operation.
• Ontario Power Generation is building a GE-Hitachi BWRX-300 at Darlington, Canada, with first power targeted around 2030.
• NuScale received US design approval, but its first project (UAMPS, Idaho) was cancelled in 2023 when estimated power costs rose to about $89/MWh after subsidies.
• Advanced designs (sodium-cooled, molten salt, high-temperature gas) are being demonstrated, and technology companies have signed power purchase agreements for future SMR output to supply data centres.

SMRs will hit their promised costs only with order books large enough for learning, and outside China that hasn’t happened yet.

Life extensions, uprates and restarts

The cheapest new low-carbon megawatt-hours from nuclear usually come from plants that already exist, by three main routes:

• Licence extensions. Most US reactors were originally licensed for 40 years; nearly all have received 20-year renewals, and a growing number have approval to operate to 80 years. France is assessing operation of its 900 MW fleet beyond 50 and 60 years. According to the IEA, extending a reactor’s life typically costs a fraction of new capacity per MWh, and often competes with new wind and solar once grid and firming costs are counted.
• Uprates. Upgrading turbines, steam generators and instrumentation can add 5–20% capacity at existing sites. US uprates since the 1970s have added the equivalent of several large reactors.
• Restarts. The Palisades plant in Michigan, closed in 2022, has been preparing to return to service with federal loan support, and Constellation plans to restart Three Mile Island Unit 1 (renamed the Crane Clean Energy Center) under a 20-year power purchase agreement with Microsoft. Japan has gradually restarted part of the fleet shut after Fukushima, once new safety reviews were done.

Early closures, on the other hand, usually raise emissions in the short term. Studies of Germany’s phase-out (for example, Jarvis, Deschenes and Jha, Journal of the European Economic Association, 2022) found the lost nuclear output was replaced largely by coal in the 2010s, with significant costs from extra air pollution. Later growth in wind and solar has since cut coal generation substantially.

How countries are deciding

• Expanding: China, India, Russia, South Korea, the UAE, Poland (first plant planned), Czechia, France (six EPR2 reactors), the UK, the US (restarts and uprates), Japan (restarting reactors).
• Phasing out or out: Germany (completed 2023), Switzerland (no new builds, under review), Spain (planned closures 2027–2035, under debate), Taiwan (last reactor shut 2025).
• Life extensions: Belgium reversed parts of its phase-out; the US is extending many licences to 80 years.

Bottom line

Nuclear power is low-carbon, safe by any per-unit measure and extremely land-efficient, and keeping existing plants running is one of the cheapest climate measures there is. Its weak points for new Western builds are cost, construction time and financial risk. On the evidence, safety and emissions are not the problem. Whether new nuclear makes sense depends on a country’s renewable resources, its financing model, and whether it can build a fleet instead of one-off megaprojects. Compare nuclear-heavy and renewable-heavy grids on the rankings and the world energy map.

Figures in this guide are typical ranges drawn from the sources above; your own numbers depend on location, prices and equipment. Use the linked calculators for site-specific estimates.