INDUSTRY TRENDS

10 min read

Nuclear Renaissance: EPR2, SMRs, and France's Energy Sovereignty

CareerOn Research

Industry & Market Intelligence DeskMarch 9, 2026

France's bet on new EPR2 reactors and small modular reactor startups is reshaping the European energy landscape, and the engineering skills it needs are already in short supply.

Evidence reviewed · August 26, 2026 · 7 sources

Updated after source review · 2026-09-2221 figures in this dossier have been withdrawn.Revision history See what was read

Executive Brief

The global energy crisis and the mandate for decarbonization have triggered an undeniable nuclear renaissance. At the forefront is France, executing a historic pivot to secure its energy sovereignty through a six-reactor EPR2 programme that EDF costed at 72.8 Md€²2020 on 18 December 2025, sited at Penly, Gravelines and Bugey, alongside investment in small modular reactors. That estimate goes to the state new-nuclear delegation for audit in the first quarter of 2026, a budget of 2.7 Md€ is validated for 2026, and the final investment decision is expected at the end of 2026. However, executing this vision hinges entirely on resolving the industry’s most critical vulnerability: a severe shortage of specialized engineering talent, project managers, and high-precision manufacturing technicians.

The Macro Context: The Global Race for Base-Load Clean Energy

For the past twenty years, nuclear energy faced political headwinds across the Western world, culminating in Germany’s complete phase-out and stagnation in the US and UK. Yet, the physics of grid stability remained unchanged: renewable energy sources like wind and solar require a reliable, low-carbon base-load. With natural gas no longer a politically viable bridge fuel in Europe, nuclear power has returned as an indispensable asset.

Globally, the market is highly competitive. China is building at scale and Russia is exporting its VVER technology; this desk prints no reactor count for either, because it opened no dated document that publishes one. In the West, the US is heavily subsidizing advanced reactor concepts via the IRA, while the UK is pushing forward with Hinkley Point C and Sizewell C. France’s strategy is not just about domestic energy production; it is about maintaining Europe’s sole domestic heavy-nuclear export capability and supply chain independence.

72.8 Md€in 2020 euros, six EPR2 reactors, EDF estimate of 18 December 2025

The operator’s own price for the programme

EDF’s published estimate, going to state audit in the first quarter of 2026. It is an estimate before the final investment decision, not an outturn cost, and it is stated in constant 2020 euros.

Source 2 Government of France

The French & European Epicenter: EPR2 and the SMR Revolution

France’s approach is dual-tracked, balancing massive centralized infrastructure with agile, venture-backed innovation.

The EPR2 Fleet: The Evolutionary Power Reactor 2 (EPR2) is designed to be simpler and cheaper to construct than its predecessor. EDF names three sites for the six reactors, Penly, Gravelines and Bugey, targets commissioning of the first reactor at Penly in 2038, and states a cadence of one reactor every twelve to eighteen months after it. The state support submitted to the European Commission on 19 November 2025 combines a below-market loan for at least half the construction cost with a forty-year contract for difference. This requires reactivating dormant supply chains and forging components at an unprecedented scale.
The Nuward Project & SMR Startups: Alongside EDF’s SMR project (Nuward), a new breed of deeptech startups has emerged from the "France 2030" initiative. Companies like Jimmy (high-temperature reactors for industrial heat), Naarea, and Renaissance Fusion are attracting venture capital to build micro-reactors designed to decarbonize heavy industry locally.

The Talent & Engineering Landscape: The "Match" Challenge

The "Plan Match" (Mettre à disposition les Talents pour le CHantier nucléaire) highlights the stark reality: the sector’s core trades held 125,000 direct jobs in 2023 inside a sector total of 220,000, and the synthesis forecasts about 60,000 full-time-equivalent recruitments of new work capacity over ten years on that core perimeter, extrapolated to about 100,000 across the whole 220,000. It states the pace this implies: up to 10,000 jobs a year.

Read the unit before the total. The synthesis counts work capacity supplied by employees who are trained, experienced and fully operational, which is why it treats the ramp as a qualification problem and not a hiring problem. A number of people is not what was measured.

For junior talent, the opportunities are vast. The industry is aggressively seeking:

Systems Engineers & Digital Twin Architects: Moving away from 2D blueprints to fully integrated 3D/4D digital modeling (using Dassault Systèmes tools).
Metallurgy & Welding Specialists: Precision welding and materials science are critical bottleneck skills for forging reactor vessels.
Supply Chain & Risk Managers: Navigating the complex geopolitical sourcing of uranium, specialized steel, and critical components.

The stigma of the nuclear industry is fading rapidly among young engineers, replaced by the realization that it is one of the most intellectually demanding and highly compensated vectors for fighting climate change.

French nuclear jobs in 2023, core trades within the whole sector (MATCH)

direct jobs, 2023, one note on one perimeter definition

Source 4 Groupement des industriels français de l'énergie nucléaire

Where the Hiring Actually Happens

The programme headline, six costed reactors at three named sites, is an investment decision. The employment reality sits in a supply chain this dossier does not size, because no dated document it opened counts the companies inside it, and that chain hires in four distinct layers. Candidates who understand the layers stop applying to the wrong one.

The owner-operator. EDF runs the fleet and the new-build programme: reactor physics, safety analysis, project control, commissioning, and the vast operations and maintenance organisation of the operating fleet. This is where nuclear careers are longest and most codified.
The fuel and component champions. Framatome (fuel, components, instrumentation and control, at Le Creusot, Saint-Marcel, Romans-sur-Isère) and Orano (conversion and enrichment at Tricastin, recycling at La Hague). Heavy forging, metallurgy, welding qualification and radiological engineering concentrate here.
Engineering and services. Assystem, Egis, Tractebel, Onet Technologies, Nuvia and Bouygues Travaux Publics supply the design offices, site works, radiation protection and maintenance outages. This layer carries the largest recruitment volume and the widest tolerance for transferable backgrounds from oil and gas, chemicals or aerospace.
The SMR cohort. EDF's Nuward, plus Newcleo, Naarea, Jimmy Energy, Otrera and Hexana. Small teams, high technical bar, real technology risk: Nuward's own 2024 decision to simplify its design toward proven components is the clearest illustration that SMR schedules are engineering hypotheses, not delivery dates.

Recruitment need over ten years to 2033, by perimeter (MATCH)

full-time-equivalent recruitments of work capacity, approximate, not a head count

Sources 3 Groupement des industriels français de l'énergie nucléaire · 4 Groupement des industriels français de l'énergie nucléaire

The Bottleneck Is Not Physicists

The MATCH synthesis puts the sector need at about 100,000 full-time-equivalent recruitments over the decade to 2033, and states where the difficulty sits. The scarcity is not in reactor physics; it is in certified industrial trades and in the engineers who can qualify them:

Welders qualified to nuclear codes (RCC-M, ASME). Flamanville 3's weld remediation campaign turned this into a board-level constraint, and the industry response, including dedicated welding academies and the "university of welding" initiatives, is now a standing recruitment channel in its own right.
Boilermakers, pipefitters and non-destructive-testing technicians. Certification takes months to years and cannot be compressed by hiring more graduates.
Quality and manufacturing surveillance. The scarcest hybrid profile in the supply chain: someone who understands both a forging process and a regulatory dossier.
Safety case and licensing engineers. Every design change must be re-argued to the regulator (ASNR since 2025, following the merger of ASN and IRSN). Documentation is not overhead here; it is the product.

Read that list again as a career instruction: the fastest route into the nuclear renaissance is a certified technical trade or a qualification-facing engineering role, not a master's in energy policy.

What this dossier will not price

This section used to carry a table of pay bands by role and seniority, seven roles and twenty-one ranges. It cited no document, and no dated document this desk could open prints remuneration by post and seniority for the French nuclear sector. The bands are withdrawn rather than re-sourced, and every withdrawn range is quoted in the revision history at the end of this dossier.

What would put a band back here is specific and checkable: a dated publication, from a ministry, a branch agreement or a named employer, printing pay by post and by seniority. Until such a series exists, the decision this dossier can actually inform is the one the qualification system governs, which posts a candidate is eligible for, in what sequence, and what evidence of code literacy and traceability discipline they can show before an employer funds a certification.

Entry Routes, Ranked by Actual Conversion

Apprenticeship inside the supply chain. EDF, Framatome, Orano and the large services firms convert apprentices at rates no external application matches, and they fund the certifications you cannot self-purchase.
Lateral move from an adjacent regulated industry. Oil and gas, chemicals, defence shipbuilding and aerospace share the vocabulary of qualification, traceability and safety cases. This is the most under-exploited door for experienced professionals.
Certified trade training. Welding, boilermaking and NDT programmes run by industry academies and the AFPA/CFA network lead to signed offers before completion in tension trades.
Graduate engineering intake. The classic route through INSTN, Grenoble INP, INSA and Centrale specialisations. Reliable, competitive, slower to responsibility than the trades.
Direct application to an SMR start-up (lowest conversion). Exciting, small, and dependent on financing rounds: a second job in nuclear, rarely a first.

What Could Go Wrong

Anyone advising a decade-long career commitment owes the downside case plainly.

Cost and schedule drift is the base rate, not the exception. Published EPR2 estimates have already been revised upward: this dossier itself printed about fifty billion euros, and EDF’s own estimate of 18 December 2025 is 72.8 Md€2020², going to audit in the first quarter of 2026. This desk prints no delay or overrun figure for Flamanville 3, because the audit report it cites was unreachable on the day of review. Programme slippage moves hiring peaks by years: it rarely cancels them, but it changes when you should move.
SMR timelines are unproven. No Western SMR is in commercial operation. Design changes, licensing novelty and financing are all live risks.
Political reversibility. Energy policy has flipped twice in fifteen years in France. Long-lived construction contracts insulate the supply chain, but not entirely.
Competition for the same people. Defence, grid renewal and data-centre construction bid for identical welders, planners and electrical engineers. That competition is good for wages and hard for programme delivery.

Is This Sector Right for You? A Short Self-Test

Nuclear rewards a specific temperament, and no salary band compensates for the wrong fit. Four honest questions.

Can you work inside a procedure and still think? Everything is documented, reviewed and traceable. People who read process as bureaucracy burn out; people who read it as engineering memory thrive.
Are you comfortable with decade-scale feedback loops? A design choice you make in 2026 is verified when a reactor operates in the late 2030s. If you need shipping velocity, this is the wrong industry.
Can you defend a judgement in writing to a hostile reader? The regulator is not an enemy, but it is adversarial by design. Written argument is the core professional skill.
Are you geographically flexible for at least three years? The work is where the concrete is. Mobility is the single largest predictor of nuclear career acceleration.

Four yes answers and scarcity compounds in your favour, because the qualifications that gate this work take months to years to earn and cannot be bought privately. Two or fewer, and adjacent sectors (grid, defence, industrial software) offer the same technical depth with a shorter cycle.

The Talent Map: Four Places the Decade Is Decided

Nuclear careers are geographic in a way software careers are not. Four locations concentrate most of the decade's new demand, and each implies a different life, not just a different job.

Penly (Seine-Maritime): the first EPR2 pair. Site preparation and early works make this the leading indicator for the whole programme. Civil works, project control, welding supervision and quality surveillance ramp first here; local housing and school pressure is already the standard complaint of relocated engineers, and it is worth investigating before signing.
Gravelines (Hauts-de-France), named with Penly and Bugey in the costed programme. The same labour pool is being bid for by battery gigafactories and port logistics, which is why nuclear employers in the region compete on training funding and permanent contracts rather than on day rates alone.
La Hague and Tricastin: the fuel cycle. Recycling, conversion and enrichment run continuously and hire independently of new-build schedules. For anyone who wants nuclear work without programme risk, this is the most stable entry point in the country.
Le Creusot and Saint-Marcel: heavy components. Forging, machining and non-destructive testing for reactor vessels and steam generators. After the quality-record incidents of the last decade, this is also where surveillance and traceability roles are most scrutinised, and therefore most valued.

The practical instruction: pick the layer and the site before polishing the CV. A generic nuclear application competes nationally; an application that names a site, a trade and a certification competes against three people.

Method and Limits

The investment figure is the operator’s own forecast estimate, not an audited outcome: EDF published 72.8 Md€2020 for the six-reactor programme on 18 December 2025 and submitted it for audit in the first quarter of 2026. An earlier figure this dossier printed is withdrawn and quoted in the revision history.
The recruitment figure is a work-capacity forecast, not a headcount. The MATCH synthesis forecasts about 60,000 full-time-equivalent recruitments of new work capacity on the 125,000 core-trade jobs, extrapolated to about 100,000 across the sector’s 220,000 jobs. It is neither 100,000 people nor 100,000 net new jobs.
This dossier prints no pay figures. No dated document it could open publishes pay by post and seniority for the sector, so the former table and the premium and tolerance printed beside it are withdrawn, and each is quoted in the revision history.
Scope. Figures describe France. Nuclear labour markets in the UK (Hinkley Point C, Sizewell C), Czechia and Poland are adjacent and currently import French expertise, generally at higher day rates.

Data & Citations

Investment: 72.8 Md€2020 forecast estimate for the six-reactor EPR2 programme at Penly, Gravelines and Bugey, with 2.7 Md€ validated for 2026 and a final investment decision expected at the end of 2026 (Source: EDF press release, 18 December 2025).
Talent Requirement: 125,000 core-trade jobs in 2023 within 220,000 sector jobs; about 60,000 full-time-equivalent recruitments of new work capacity forecast over ten years on the core perimeter, extrapolated to about 100,000 across the sector, at a pace of up to 10,000 jobs a year (Source: GIFEN, programme MATCH synthesis, April 2023).
Global Context: In 2022 nuclear supplied 62.8% of the electricity generated in France and 21.8% of EU electricity production (Source: Eurostat, 12 January 2024).

The argument

The spine of this dossier

Every part below rests on citations this dossier carries, including the part that says what would prove us wrong.

The question

A reactor programme publishes a recruitment total and an investment total. Why does neither number tell a candidate when to move, and what in the same documents does?

One decision a named reader faces, phrased as a question. A topic cannot be answered, so it cannot be wrong.
The tension

The two most quoted numbers about this programme measure something other than what they are quoted for. The sector recruitment total is a forecast of full-time-equivalent work capacity, supplied by employees the same synthesis describes as already trained, experienced and fully operational, not a count of people to be hired. The investment total is the operator’s own forecast estimate, submitted for audit rather than audited, and the earlier figure this dossier itself printed has since been replaced by its own publisher. A total that is neither a headcount nor an audited cost cannot date a hiring decision.

Two things in the published record that do not fit, each citing a document this dossier carries.

Citations [2] [3]
The mechanism

Each publisher measures the act it is obliged to record, which is why the unit drifts from the thing a reader wants. A sector body must forecast capacity to plan training, so it counts work capacity and states the qualification condition attached to it; an operator must state a forecast cost to obtain a decision, so it publishes an estimate with an audit date rather than an outcome. Therefore the usable content of both documents is not the total but the schedule and the condition inside it: named sites, a commissioning target, a cadence between reactors, an audit quarter, a decision date, and a pace of jobs per year.

Why the tension exists: the causal step, stated so it can be disputed on its own terms.

Citations [2] [3]
What follows

For a candidate the consequence is that the qualification comes before the vacancy, because the forecast counts capacity that is already qualified. The fastest route into this programme is therefore a certified technical trade or a qualification-facing engineering role, entered through an employer that funds the certification, and the timing is read off the operator’s dated schedule rather than off a national total. An engineer or technician who arrives with codes, traceability and a written safety argument is the unit the forecast is actually short of.

What follows for the reader: which skill gets priced, which comparison stops working, which offer is not what it looks like.

Citations [2] [3]
What would prove us wrong

The reading fails where the sector body publishes a later edition stating the total as persons hired, gross of retirements, and withdraws the work-capacity unit. It fails equally where the audit of the operator’s estimate is published and confirms the estimate unchanged while the commissioning target and the cadence move, which would make the total the stable number and the schedule the unreliable one.

The observation that would show this argument is wrong, and the source that would carry it.

Citations [2] [3]
Our position

The energy desk’s reading: this dossier prints the operator’s forecast estimate with the day it was published and the audit it goes to, prints the sector forecast in the unit its publisher used, and prints no pay figure at all, because no dated document we could open publishes remuneration by post and seniority for this sector. The bands this dossier once carried, the premium and tolerance beside them, an uncited supplier count, a site count, a delay and an overrun attributed to a report we could not open, and a reactor count for another country are all withdrawn and quoted in the revision history rather than deleted.

What this desk concludes, signed and dated, resting only on the citations named above.

Citations [2] [3] [6]

Signed by energy-desk, 2026-09-22

Every argument in the corpus

Shape the future of career exploration

Want a simulation for this role or industry?

We're building a library of job simulations shaped by the roles readers ask for. Tell us what you'd like to practice, and we'll work with our partners to create it, completely free.

Request a Simulation

Revision history21 changes · 2026-09-22

Withdrawn figures

Figures this dossier printed and has withdrawn

Each figure below appeared in an earlier version of this dossier. It was withdrawn on the day stated, after the document it was attributed to was read. Where that document prints a value, the value is restated beside it and cites this dossier’s own source list.

The publisher revised the figure · 2026-09-22

Estimated €51.7 billion for the first six EPR2 reactors

Said to measure: the published cost of the six-reactor EPR2 programme

Attributed to: EDF, Credited in this dossier as "EDF/French Government audits", with no edition named

What was read: EDF published its own forecast estimate for the six-reactor EPR2 programme on 18 December 2025 at 72,8 Md€2020, for Penly, Gravelines and Bugey, and stated that the estimate goes to the state programme delegation for audit in the first quarter of 2026. The figure this dossier printed is the publisher’s superseded number, and the credit beside it named no edition that could have dated it.

Restated as: 72.8 Md€2020 for six EPR2 reactors (EDF, 18 December 2025) (citation [2])
The publisher revised the figure · 2026-09-22

The €51.7bn figure reflects an earlier EDF estimate

Said to measure: the method note conceding that more recent estimates were higher without naming one

Attributed to: EDF, Method note printed in this dossier, citing no edition

What was read: The note said the later estimates were higher and left the reader without one, which keeps a superseded figure in circulation while admitting it is superseded. The dated estimate existed and is now printed in the dossier instead.

Not restated: no located document prints a value.
The publisher revised the figure · 2026-09-22

€50+ billion infrastructure mega-project

Said to measure: the order of magnitude of the programme, printed in the opening brief

Attributed to: No publisher: the brief carried no source, Executive brief printed in this dossier with no cited document

What was read: A rounded floor for a cost its publisher has since stated precisely is not a simplification, it is a looser version of a number that exists. The brief now prints the publisher’s own estimate and the date it carries.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

at least six (and potentially up to fourteen) next-generation EPR2 reactors

Said to measure: the size of the reactor programme a candidate was being asked to plan a decade around

Attributed to: No publisher: the sentence carried no source, Executive brief printed in this dossier with no cited document

What was read: The costed and sited programme is six reactors: EDF names Penly, Gravelines and Bugey on 18 December 2025 and puts the final investment decision at the end of 2026. The further eight are a political ambition no document read for this dossier commits to, and a candidate reading a hiring ramp cannot tell the two apart when they are printed in one sentence.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

six EPR2 reactors, with an option on eight more

Said to measure: the programme headline used to introduce where the hiring happens

Attributed to: No publisher: the sentence carried no source, Section opening printed in this dossier with no cited document

What was read: An option is a contractual object and no document read for this dossier publishes one for eight reactors. What EDF publishes is a costed six-reactor programme at three named sites, a cadence of one reactor every twelve to eighteen months, and a decision date.

Not restated: no located document prints a value.
The figure measures something else · 2026-09-22

must recruit roughly 100,000 new professionals by 2033

Said to measure: the headline talent requirement of the French nuclear programme

Attributed to: GIFEN, Programme MATCH: synthèse de la note remise au Gouvernement (avril 2023)

What was read: The synthesis does not count professionals. It forecasts about 60,000 full-time-equivalent recruitments of new work capacity across the 125,000 core-trade jobs the sector held in 2023, and extrapolates about 100,000 across the 220,000 jobs of the whole sector. A full-time-equivalent of new work capacity is not a person hired, and the document says so in the same paragraph, adding that these are people already trained, experienced and fully operational.

Restated as: about 60,000 full-time-equivalent recruitments of new work capacity on the 125,000 core-trade jobs, extrapolated to about 100,000 across the sector’s 220,000 jobs (citation [3])
The figure measures something else · 2026-09-22

counts gross recruitments over a decade, including replacement of retirements

Said to measure: the method note explaining what the 100,000 figure measures

Attributed to: GIFEN, Programme MATCH: synthèse de la note remise au Gouvernement (avril 2023)

What was read: The note was a second misreading laid over the first: the synthesis prints "capacité de travail nouvelle", new work capacity, and nowhere frames the total as gross hiring including retirements. The distinction matters to a reader deciding when to move, because a replacement need and a capacity need peak in different years.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

a supply chain of roughly 3,200 French companies

Said to measure: the size of the employer population a candidate could apply to

Attributed to: No publisher: the sentence carried no source, Section opening printed in this dossier with no cited document

What was read: No dated document read for this dossier counts the companies in the French nuclear supply chain. The MATCH synthesis counts jobs and perimeters, not firms, and the sector observatory page it sits behind publishes no edition. The count leaves the dossier and the four hiring layers, which are named employers a reader can check, carry the section instead.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

across 18 sites

Said to measure: the number of sites across which the owner-operator runs operations and maintenance

Attributed to: No publisher: the sentence carried no source, Hiring-layer list printed in this dossier with no cited document

What was read: None of the seven documents this dossier cites prints a site count for the operating fleet, and the sentence credited nothing. The layer is described by what it does instead, which is the part a candidate applies to.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

China is currently constructing over 20 reactors

Said to measure: the scale of competing construction programmes outside Europe

Attributed to: No publisher: the sentence carried no source, Macro-context paragraph printed in this dossier with no cited document

What was read: The dossier’s only statistical source is Eurostat, which counts European production and reactors, not construction elsewhere. A count with no document behind it adds nothing a reader can act on, and the paragraph’s point, that heavy-nuclear capability is contested, stands without it.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

commissioned roughly twelve years late at several times its initial budget

Said to measure: the base rate of cost and schedule drift a candidate should price into a decade-long commitment

Attributed to: Cour des comptes, La filière EPR : une dynamique nouvelle, des risques persistants (cited without a page or an edition)

What was read: The figures were printed as common knowledge beside a report the dossier cites but never quotes, and the report was not reachable on the day of review, so nothing could be located to carry them. The risk the bullet exists to state is now made from the documents this dossier can open: its own superseded cost figure, EDF’s 72,8 Md€2020 estimate of 18 December 2025, and the audit of that estimate due in the first quarter of 2026.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

Outage and mobility premiums can add 5-20% in field roles

Said to measure: the premium a field role adds to the base pay bands printed beside it

Attributed to: APEC in the French mirror, nobody in the English, Credited in the French text only as "repères APEC, 2024", a document with no title, no edition and no address

What was read: The two languages did not even agree on whether the range had a source: the French credited an undated APEC reference the English did not print. No document this desk could open publishes an outage or mobility premium range for the French nuclear sector, so the range is withdrawn rather than re-credited.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

negotiation anchors with roughly ±10% error

Said to measure: the precision claimed for the withdrawn pay bands

Attributed to: No publisher: the note carried no source, Method note printed in this dossier with no cited document

What was read: An error margin is a property of a measurement, and there was no measurement: the bands were a synthesis of unnamed postings and recruiter ranges. Printing a tolerance around them borrowed the authority of a statistic none of them had.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

the horizon is a 60-year asset life

Said to measure: the career horizon offered in exchange for base pay below software

Attributed to: No publisher: the sentence carried no source, Compensation paragraph printed in this dossier with no cited document

What was read: A design life is an engineering and licensing figure, granted reactor by reactor, and no document read for this dossier prints one for EPR2. What EDF does print is a commissioning target of 2038 for the first reactor at Penly and a cadence of twelve to eighteen months thereafter, which is the horizon a candidate can actually plan against.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

28 000-36 000 €

Said to measure: the entry pay band for a certified nuclear welder in France

Attributed to: No publisher: the table carried no source, Compensation Landscape table printed in this dossier with no cited document

What was read: The table was published as a benchmark with three bands for each of seven roles and cited nothing. No dated document this desk could open prints pay by post and seniority for the French nuclear sector, so all twenty-one ranges are withdrawn rather than re-sourced.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

26 000-34 000 €

Said to measure: the entry pay band for a maintenance and outage technician

Attributed to: No publisher: the table carried no source, Compensation Landscape table printed in this dossier with no cited document

What was read: Outage work is contracted through the services layer, where day rates and permanent pay differ by employer and by campaign. A single national range for it could only have been an average of documents nobody named.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

36 000-44 000 €

Said to measure: the entry pay band for a mechanical or systems engineer

Attributed to: No publisher: the table carried no source, Compensation Landscape table printed in this dossier with no cited document

What was read: The row was the one most likely to be quoted back to an employer, and it rested on nothing an employer publishes. Neither EDF nor Framatome nor the engineering firms named in this dossier print a band for the role.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

38 000-46 000 €

Said to measure: the entry pay band for a safety case and licensing engineer

Attributed to: No publisher: the table carried no source, Compensation Landscape table printed in this dossier with no cited document

What was read: The dossier argues that written argument is the core professional skill in this sector, and then priced that skill with an unsourced number. The argument survives the withdrawal; the price did not deserve to.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

34 000-42 000 €

Said to measure: the entry pay band for a radiation protection engineer

Attributed to: No publisher: the table carried no source, Compensation Landscape table printed in this dossier with no cited document

What was read: Radiation protection pay is set inside regulated employers and their branch agreements, none of which this dossier cites or read. The band had the shape of a document and the substance of an estimate.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

35 000-43 000 €

Said to measure: the entry pay band for project control and planning

Attributed to: No publisher: the table carried no source, Compensation Landscape table printed in this dossier with no cited document

What was read: Project control is the role most often filled from adjacent industries, so its pay is the most employer-specific of the seven and the least reducible to one national range.

Not restated: no located document prints a value.
The document does not exist · 2026-09-22

37 000-45 000 €

Said to measure: the entry pay band for an instrumentation and control engineer

Attributed to: No publisher: the table carried no source, Compensation Landscape table printed in this dossier with no cited document

What was read: The last row of the table closes the same hole as the first: seven roles, three seniorities, twenty-one ranges, and not one document behind any of them. The dossier now states what it will not price and what would let it price it.

Not restated: no located document prints a value.

Read the full withdrawal register

Industry Atlas

Part 2 of 13

France 2030 and the deeptech bet: how state-funded science actually hires

Launch Pad to Career: France's New Space Revolution and Ariane 6