Nuclear encyclopedia

Chapters / The economics of nuclear power

The economics of nuclear power

Why nuclear plants cost what they cost, why Western projects keep blowing budgets, and how nuclear compares on system costs.

Status: cost benchmarks and three case files published; financing models to come. Release v0.8.1, 20 November 2026. Bracketed [source-id] links point to the source register.

Nuclear power is cheap to run and ruinously expensive to build. That single fact explains most of its history: the technology wins where capital is cheap, construction is repeatable, and the state absorbs risk; it struggles where every project is a first-of-a-kind megaproject.

Comparisons of levelised cost of electricity routinely place new nuclear above wind and solar in the West and below them in parts of Asia, but LCOE arguments miss the system question: what it costs to keep the lights on when the sun is down. This chapter keeps the accounting visible and cites every number.

The shape of the cost problem

A nuclear plant inverts the usual utility bill: the great majority of lifetime cost is capital spent before the first megawatt-hour, while fuel is a few per cent of generation cost. That makes the economics exquisitely sensitive to construction time and the cost of capital, a point the OECD Nuclear Energy Agency's 2025 cost study quantifies across discount rates: double the financing rate and a nuclear project's LCOE climbs far more steeply than a gas plant's. It also means an operating plant, its capital sunk, is often one of the cheapest sources on the grid. [nea-costs-2025]

The benchmark numbers

Lazard's LCOE+ (June 2025) puts new-build US nuclear at $141-220 per MWh, against $37-86 for onshore wind and $38-78 for utility solar, on capital costs of $9,020-14,820 per kW, an 84-month build and a 70-year life assumption. Two caveats belong with that range: Lazard bases its nuclear estimate on the Vogtle experience, a first-of-a-kind US project, and LCOE measures the plant, not the system. Firm capacity, grid services and overbuild requirements are priced elsewhere. [lazard-lcoe-2025]

Case file: Vogtle

The first US reactors built from scratch in three decades, Vogtle units 3 and 4, were estimated around $14 billion when construction began in 2009 and finished above $35 billion; unit 3 entered service in July 2023 and unit 4 in April 2024, roughly seven years late. The project is the standard exhibit for first-of-a-kind risk, and also for the learning curve: Lazard estimates about a 30 per cent cost improvement between unit 3 and unit 4 alone. [reuters-vogtle-cost] [lazard-lcoe-2025]

Case file: the EPR in Europe

France's Court of Accounts put Flamanville 3 at €23.7 billion (2023 euros) including financing, against a €3.3 billion construction estimate in 2006, with grid connection twelve years late. Britain's Hinkley Point C, two EPRs once promised for 2017 at £18 billion, was re-estimated in January 2024 at £31-34 billion in 2015 prices with first power in 2029-2031. Both are first-of-a-kind European builds after a two-decade construction gap, and both projects argue their lessons transfer to the follow-on units now planned. [ccomptes-epr] [edf-hpc-2024]

Case file: Barakah, the counterexample

The UAE signed a $20 billion contract with a KEPCO-led consortium in December 2009 for four APR-1400 reactors at Barakah. The units entered commercial operation in 2021, 2022, 2023 and September 2024: 5.6 GW supplying about a quarter of the country's electricity, delivered close to the original contract structure by a team repeating a proven Korean design at fleet scale. South Korea's own build record, with costs roughly half of Western levels and construction times near five years, underlies it. [enec-barakah] [pt-barakah]

Operating economics: why old plants are loved

Once built, nuclear plants run at capacity factors around 90 per cent (the US fleet's long-run figure) with fuel a minor cost line, which is why licence extensions to 60 and 80 years dominate the US fleet and why post-2022 Europe moved from phase-outs to lifetime extensions. The cheap electron is the amortised one. [lazard-lcoe-2025] [iaea-rds2-46]

Still to come in this chapter

  1. Financing models in depth: regulated asset base, contracts for difference, state builds
  2. Decommissioning and waste funds: who pays, and is it enough
  3. LCOE versus system cost: what grids actually pay for firm low-carbon power
  4. South Korea and China build-cost records, with primary sources

Claims in the final text will carry citations to the source register; anything unverifiable will land on the gaps page instead.