Executive brief
Two published financial years put a mechanism in plain sight that no defence headline describes. In its FY2024 publication Thales reported order intake of €25.3bn and sales of €20.6bn; in its FY2025 publication it reported order intake of €25.3bn again and sales of €22.1bn, with adjusted EBIT rising from €2.419bn to €2.740bn (Thales, FY2024 and FY2025 results). Intake was flat. Delivery was not. And the recruitment plans published on either side of that pair moved the same way as delivery, not as intake: 8,000 hires worldwide including 3,000 in France announced for 2025, and 9,000 worldwide including 3,300 in France for 2026 (Thales, 2025 and 2026 recruitment announcements). The career-relevant reading is therefore not "orders are booming" — by the company's own two numbers they did not grow at all — but that a European defence prime hires against conversion of a backlog it already holds. That distinction decides which roles open, in which order, and how long they last. This dossier prints the figures at the level they were published, refuses every site-level headcount and programme value we could not source, and states what it does not know.
I. The mechanism: a backlog is a hiring instruction with a delay
Defence coverage treats order intake as the signal. For an engineer choosing an employer it is close to the wrong number. An order signed in one year is an obligation to deliver across several, and delivery is what consumes hours: integration hours, test hours, qualification hours, industrialisation hours. So the recruitment curve of a prime tracks the conversion of a book already won, and it keeps tracking it even in a year when intake is flat.
Thales's own two published years show precisely that shape. Order intake was €25.3bn in the FY2024 publication and €25.3bn in the FY2025 publication. Sales rose from €20.6bn to €22.1bn, adjusted EBIT from €2.419bn to €2.740bn, and adjusted net income from €1.9bn to €2.005bn (Thales, FY2024 and FY2025 results). A reader who indexed their career decision on intake would have concluded that 2025 was a standstill year. The delivery line says it was the opposite.
Thales, as published: two financial years side by side (FY2024 and FY2025 results releases)
| Measure | FY2024, as published | FY2025, as published |
|---|---|---|
| Order intake | €25.3bn | €25.3bn |
| Sales | €20.6bn | €22.1bn |
| Adjusted EBIT | €2.419bn | €2.740bn |
| Adjusted net income | €1.9bn | €2.005bn |
Sources 1 Thales Group · 2 Thales Group
II. What the recruitment plans actually say — and at what level
Two dated recruitment statements bracket that pair of years. For 2025 the group announced a plan of 8,000 hires worldwide, of which 3,000 in France. For 2026 it announced 9,000 worldwide, of which 3,300 in France (Thales, 2025 and 2026 recruitment announcements). These are year-specific plans, and we treat them as such: they are not added together, not annualised into a trend line, and not read as a headcount forecast. A hiring plan is an intention published by an employer; an executed hire is a different fact, and the group publishes the second only as a total workforce figure.
3,300
hires planned in France for 2026, inside a worldwide plan of 9,000, as announced by the company
A published plan for one named year, not an outcome and not a run rate. The comparable 2025 plan was 3,000 in France inside 8,000 worldwide. We do not sum the two years, and we publish no executed-hire count for either.
Sources 3 Thales Group · 4 Thales Group
On total workforce the record is already ambiguous, and the ambiguity is worth a candidate's attention. The group describes itself as employing more than 85,000 people; a third-party company register carried 86,043 for a comparable period. Those are two different acts of publication — a self-description with a floor, and a register entry with a precision it does not justify — and we print both rather than choosing the tidier one.
III. Where the conversion work sits: programmes, not press releases
Three published programme facts explain the shape of the demand without requiring a single number we cannot source.
Ground-based air defence. The company has announced serial production of SAMP/T NG, the new-generation surface-to-air system, describing the work in terms of seven sections of the system rather than a euro value; no contract value is published, and we print none (Thales, SAMP/T NG announcement).
Military satellite communications. The Syracuse IV generation and its associated ground segment, announced with a scope of 30 stations, is likewise published without a disclosed value (Thales, Syracuse programme announcement). Two programmes, two published scopes, zero published prices: that is a normal condition of the sector and it is why compensation and programme economics cannot be inferred from the news.
The uncertain one. Coverage of the Future Combat Air System has repeatedly reported friction between the programme's industrial partners and doubt over its schedule (Reuters, 2025). We do not resolve that; we record it, because a candidate weighing a fifteen-year programme against a system already in serial production is making a risk choice, not a prestige choice.
There is also a research edge worth naming precisely because it is small. The group's quantum-sensing work appears in the public record through collaborative projects — including one described with a total project value of €2.6m — a figure two orders of magnitude below the programmes above. Read as a career signal it says: real, funded, and thin. It is a laboratory route, not a volume employer.
IV. The skills the record actually supports
Nothing in what we could open licenses a ranked list of "hot skills". What the published record does support is a reading of which kind of engineering converts a backlog into revenue at a prime:
- System integration and verification. The unit of delivery is a qualified system, not a component. Sections of an air-defence system and stations of a satcom ground segment are integration objects.
- Radar, RF and mission electronics. The programmes named above are electronics programmes with an airframe or a vehicle attached, not the reverse.
- Software and cyber inside a certified process. The group's own public positioning ties defence electronics to cyber and to secure digital infrastructure; the constraint on that work is accreditation, not framework fashion.
- Industrialisation. Serial production of a system previously built in prototype quantities is a manufacturing-engineering problem, and it is the phase a flat-intake, rising-delivery year is in.
And one structural gate that no other employer class imposes so hard: security clearance. It is granted by a state, to a person, for a scope of work. It cannot be acquired in advance, it constrains nationality and mobility, and it is the single most consequential non-technical variable in a defence career. Any candidate should establish, in writing, which of their target roles requires clearance and who sponsors the request.
V. What we refuse, and why the refusal is the useful part
We could not verify, and therefore do not publish, the headcount of any individual French Thales site. Figures for Brest and Cholet circulate in regional reporting; we could not open a company publication that states them, and a site headcount is exactly the kind of number that gets copied between outlets until it looks primary. Several of the group's own pages refused our fetcher outright, and where that happened we relied on the released figures as published rather than quoting a page we did not read.
We also refuse the number a reader most wants: the share of group revenue attributable to defence. No consolidated figure we could verify exists in the sources we opened, and constructing one from segment tables would be our arithmetic presented as the company's disclosure.
VI. The conclusion the evidence supports
On two published financial years, Thales is a prime whose intake was flat and whose delivery, profit and hiring plans all rose. That is the profile of a company working through a book, not chasing one — and it is a better environment for an engineer who wants to be handed a system to qualify than for one who wants to be near the signature of a new contract. The programmes with published scope are in serial production or in ground-segment rollout; the programme with the most publicity is the one with the most reported schedule doubt; and the entry condition that decides everything is a clearance, not a keyword.
Refused in print
We do not publish, and did not estimate: any headcount for Brest, Cholet, Toulouse, Mérignac, Gennevilliers or any other individual site; any defence share of group revenue; any contract value for SAMP/T NG or for the Syracuse IV generation, neither of which is published; any executed-hire count for 2025 or 2026, as opposed to the announced plans; any salary, salary band or compensation figure for any role; any merged multi-year hiring total; any FCAS schedule, workshare or budget; and any figure quoted from a company page that refused our fetcher.