France produces electricity among the most decarbonized in the world, yet risks running short of reactors to get through the next decade. Not by accident, but because the two programs supposed to guarantee continuity of the fleet (the grand carénage to extend the old ones, the EPR2 to build new ones) are accumulating delays and cost overruns while the average age of reactors in service approaches 40 years. The 2027 decision is not about choosing between nuclear and renewables. It is about deciding who pays what, according to which rules, to prevent the industrial base from collapsing before its successor arrives.
Forty-year-old reactors, and a gaping hole in the schedule
French nuclear production peaked around 2005 and 2006, at nearly 430 TWh per year [1]. In 2022, at the height of the stress corrosion cracking crisis, it fell to 279 TWh [1]. It has recovered since: 361.7 TWh in 2024, 373 TWh in 2025 [2, 3]. This rebound is real, but it masks a deeper structural problem.
France’s 56 reactors were built essentially between 1978 and 1994, in less than twenty years, during an industrial effort without parallel. In 2024, their average age reached 39 years, nearly the initial lifespan for which they were designed. A cliff effect is expected in the 2040s, with the closure of many power plants built in the 1980s: similar to a quarter of global nuclear capacity that will have to be shut down at that horizon.
EDF launched the grand carénage to extend these reactors. Starting from an estimate of 45 billion euros in 2018, the program had already been revised to 49.4 billion in 2020 [4]. But the Court of Auditors, in its January 2025 report, valued the total cost of the grand carénage at 100.8 billion euros over the period 2014-2035, and even 131.9 billion when including routine maintenance. The shift from 40 to 60 years of lifespan, now the official target, costs roughly 51 euros per megawatt-hour produced, against 35 euros in the previous assessment [5]. Power plants age, and each decennial inspection reveals new requirements.
The Flamanville EPR as an industrial lesson
The first EPR, at Flamanville, illustrates what an initial budget is worth in the nuclear sector. Launched in 2007 with a budget of 3.3 billion euros, it ultimately cost 23.7 billion euros according to the Court of Auditors, more than seven times the initial budget. The project was completed in autumn 2024 instead of June 2012, meaning over two hundred months of construction instead of the fifty-four planned: a delay multiplied by 3.8.
This is not an isolated case: the Finnish EPR reactor experienced the same twelve-year delay [6]. What changed is that the industry lost its expertise during the twenty years between Flamanville and the decision to restart nuclear power. Connected to the grid in December 2024 with a 12-year delay, this project symbolized the loss of expertise in the industry and the cost overruns of French nuclear.
EPR2: three years of delay accumulated before the first concrete is poured
The program for six EPR2 reactors, announced by Emmanuel Macron in 2022, was supposed to have its first reactor in service between 2035 and 2037. The commissioning of the first EPR2 reactor at Penly is now expected in 2038. Three years of slippage, and the final investment decision has not yet been made.
Late 2025, EDF reassessed the cost of the six future EPR2s at 72.8 billion euros, an 8% increase. But this amount is in 2020 euros and excludes financing costs. The Court of Auditors, in its January 2025 report, mentions a bill of 79.9 billion euros at 2023 prices. Its president Pierre Moscovici mentioned a cost “likely to exceed 100 billion euros” once financing costs are included.
At the stage of this report’s publication, the EPR2 program is characterized by a design still to be firmed up, an incomplete budget, and financing to be clarified. This is not a doomsday assessment: it is the Court of Auditors’ evaluation, issued in January 2025, after two critical reports published over the decade [7]. The nuclear industry is “far from ready” to carry out the French new nuclear program and faces “concerning” challenges.
A financial structure has nevertheless taken shape since. The scheme submitted to Brussels comprises a subsidized state loan covering 60% of the total program amount, financed by the savings fund of the Caisse des dépôts, and a contract for difference over forty years, intended to offset differences between a reference price and the market price. This contract for difference guarantees EDF a capped price of 100 euros per megawatt-hour [8]. In other words: if the market price falls below that threshold (which will be the case most of the time), the state pays the difference for forty years. The risk of cost overruns remains partially pooled.
The European Commission opened a consultation on this structure in May 2026. Its validation conditions the actual launch of construction [9].
The human constraint, harder even than the financial constraint
The industrial capacity to build these reactors poses a problem as concrete as the financing question. The French nuclear industry plans to hire approximately 100,000 people by 2033 to support the EPR2 projects, the extension of the existing fleet, and small modular reactor programs. These are welders, boilermakers, site engineers: professions that take years to train, not months.
The industry employs 247,000 full-time equivalents across 1,830 companies, and it continues to need to hire 100,000 people in ten years, even as it is already recruiting 10,000 people per year for the past two years. Half of these hires serve to replace departures; the other half creates new positions. The pace is insufficient to manage both the grand carénage and the start of EPR2s.
Jean-Marc Jancovici and the Shift Project have documented this tension: “French low-carbon electricity production must increase sharply to allow electrification” of uses: heating, transport, industry. If electrification is the engine of decarbonization, the capacity to produce controlled current in sufficient volume is the condition of that decarbonization. Nuclear ensures its continuity where renewables do not guarantee availability in calm and cold weather.
The paradox of the moment, documented by RTE’s 2025 Provisional Assessment, is that France is currently in a situation of electricity overcapacity, with an expectation of failure now below 0.1 hours per year. Between 2027 and 2030, margins for managing winter peak demand could progressively tighten, leading to a need for additional capacities in a rapid decarbonization pathway. The window is narrow.
What the current architecture does not resolve
The issue is not the principle of a nuclear restart. Since the Belfort speech of 2022, political consensus has been reached on this point. The issue is the financing and governance architecture that frames it, and which determines who bears the risk when (as history demonstrates) schedules and costs slip.
The current structure places the state as lender of 60% of construction costs, and as guaranteed buyer of electricity produced for forty years at a floor price. EDF remains sole project owner, without competitor or external incentive. This scheme transfers program risk to the taxpayer while leaving entire industrial control to the operator whose management failures produced the Flamanville overrun.
Work by the Court of Auditors [7] and analyses by Benjamin Frémaux on financing the energy transition [10] converge on the same finding: the current financing model confuses necessary public support to make the program possible with the absence of counterparties in terms of industrial governance and risk-sharing.
The architectural choice for 2027
The final investment decision on EPR2 is expected by end 2026. It will be the most binding political act of the presidential term, well before the election. What is at stake in 2027 is the nature of the contract framing this investment, not its principle.
The lever that would change the architecture (and not merely adjust the parameters of the contract for difference) consists of establishing multi-stakeholder governance of project management for the EPR2 program, separating public financing from industrial oversight. Concretely: the state lends at a subsidized rate and guarantees a sales price, but project management associates an independent oversight committee with a right to suspend in case of observed overrun, modeled on audits by the Nuclear Safety and Radiation Protection Authority (ASNR): and financing of each subsequent tranche is conditional on validation of the actual cost and schedule of the previous tranche. This principle of conditional milestones was explicitly recommended by the Court of Auditors in January 2025 [7].
After this break, project management would no longer function as a closed in-house job site: the state would no longer be simply a passive lender of a program whose schedules and costs it does not control, but co-decision-maker at each milestone. This shift (in who decides at each phase, and on what basis) is the only way to break with the “too big to fail” logic that produced the Flamanville overrun and that the Court of Auditors, in its 2025 report, identifies as the main threat weighing on the EPR2 program [7].
Sources
[1] Connaissance des énergies, “Electricity production in France: a very low-carbon mix,” annual data 2000-2023, https://www.connaissancedesenergies.org/la-production-delectricite-en-france-un-mix-tres-peu-carbone (accessed 30/09/2026).
[2] RTE, “2024 Electricity Report: Complete Report,” April 2025, https://assets.rte-france.com/prod/public/2025-04/2025-04-09-bilan-electrique-2024-rapport-complet.pdf (accessed 30/09/2026).
[3] RTE, “2025 Electricity Report: Main Results,” February 2026, https://assets.rte-france.com/prod/public/2026-02/Bilan-electrique-2025-principaux-resultats.pdf (accessed 30/09/2026).
[4] EDF, press release, “Revision of Grand Carénage Cost,” October 29, 2020, https://www.connaissancedesenergies.org/afp/494-milliards-deuros-edf-revoit-la-hausse-le-cout-du-grand-carenage-de-son-parc-nucleaire-201029 (accessed 30/09/2026).
[5] Opera Energie, “EDF: The Court of Auditors Considers Reactor Extension Beyond 60 Years ‘Profitable,’” 2026, https://opera-energie.com/media/edf-la-cour-des-compte-estime-rentable-la-prolongation-des-reacteurs-au-dela-de-60-ans/ (accessed 30/09/2026).
[6] Franceinfo, “Why Did EDF So Underestimate the Cost and Duration of the Flamanville EPR Project?,” December 2024, https://www.franceinfo.fr/environnement/energie/nucleaire/pourquoi-edf-a-t-il-autant-sous-estime-le-cout-et-la-duree-du-chantier-de-l-epr-de-flamanville_6829382.html (accessed 30/09/2026).
[7] Court of Auditors, “The EPR Industry: New Momentum, Persistent Risks,” thematic public report, January 14, 2025, https://www.ccomptes.fr/sites/default/files/2025-01/20250114-synthese-La-filiere-EPR–une-dynamique-nouvelle-des-risques-persistants_0.pdf (accessed 30/09/2026).
[8] EDF / SFEN, “EPR2 Program: A Ceiling of 72.8 Billion Euros for Six First Reactors,” December 2025, https://www.sfen.org/rgn/programme-epr2-un-plafond-a-728-milliards-deuros-pour-les-six-premiers-reacteurs/ (accessed 30/09/2026).
[9] SFEN, “State Aid, CfD, Public Loan: Brussels Clarifies Financing for EPR2 and Opens European Consultation,” updated May 29, 2026, https://www.sfen.org/rgn/aides-detat-cfd-pret-public-bruxelles-detaille-le-financement-des-epr2-et-ouvre-une-consultation-europeenne/ (accessed 30/09/2026).
[10] RTE, “2025-2035 Provisional Assessment: Main Results,” December 2025, https://assets.rte-france.com/prod/public/2025-12/2025-12-16-bilan-previsionnel-principaux-resultats-2025.pdf (accessed 30/09/2026).
[11] Jean-Marc Jancovici / The Shift Project, “Robust Plan for the French Economy,” 2026, https://www.revolution-energetique.com/actus/voiture-electrique-train-nucleaire-les-20-projets-de-jean-marc-jancovici-pour-decarboner-la-france/ (accessed 30/09/2026).
[12] GIFEN / France Travail, “Nuclear is Hiring: 100,000 Talents Sought by 2035,” 2026, https://www.francetravail.org/accueil/communiques/2026/le-nucleaire-recruite-100-000-talents-recherches-d-ici-a-2035.html (accessed 30/09/2026).
[13] Opera Energie, “EPR2: 6 Crucial Months for French Nuclear,” 2026, https://opera-energie.com/media/epr2-6-mois-decisifs-pour-le-nucleaire-francais/ (accessed 30/09/2026).



