A report from the European sector’s annual conference, held in 2026, states a finding that sums up the shift under way: space-based Earth observation was described as having moved from a primarily civilian and environmental capability to a core component of European security and defence.
That shift is not merely thematic. It changes technical requirements, purchasing channels and market access conditions, and it opens for a data provider a position civilian uses do not present.
This article describes that evolution, the European institutional architecture and what the field requires of a provider. It extends the article on Earth observation and climate.
The separation between civil and defence in Earth observation
This legal distinction structures the whole European field and it is poorly understood outside it.
An institutional analysis recalls that for legal reasons, EU space programmes are formally civilian but may support Common Security and Defence Policy missions if member states decide.
That prudent legal formulation covers a very real operational constraint. The conference report cited above states that at the application level, Earth observation capabilities often cannot be simultaneously civilian and military in practice, the European flagship programme having been cited as an example where the civilian mandate prevented its use in a defence context, which in turn prompted the development of a separate governmental service capability.
That observation matters and it strongly qualifies the common discourse on dual use. The data may be technically identical; the regime of employment is not.
The European Earth observation architecture
How its components articulate deserves setting out.
An official publication describes that architecture: Earth observation supplies free and open data on land, oceans, the atmosphere, climate change, emergency management and security; satellite navigation supplies positioning services; and protection and secure communications rest on space situational awareness, governmental communications and a new connectivity constellation.
That last component saw recent acceleration. The European Commission states that on 7 August 2026 the Commission and the selected consortium signed an implementation agreement expanding the main constellation to 348 satellites and moving the programme from planning to full-scale deployment.
The same source specifies that the agreement confirms the timetable for developing and deploying the satellites, procuring launch services, building the secure ground segment and providing connectivity services, the expanded constellation comprising 330 satellites in higher low Earth orbit and 18 in medium Earth orbit.
That constellation concerns connectivity rather than imagery, and it nonetheless matters to the field: it constitutes the secure transmission infrastructure on which the exploitation of sensitive data will rest.
The governmental Earth observation service
This initiative concentrates the sector’s attention.
A specialist publication reports that an initiative proposed by the European Commission for the budget covering 2028 to 2034 would provide all EU members with remote sensing data as a service, a concept first laid out in 2024 with the launch of two year-long feasibility studies and a pilot project.
The same source sets the context: only a handful of EU countries, notably France, Germany, Italy and Spain, currently operate their own military satellites for remote sensing, and the collective European network is aimed at civil uses rather than military.
The conference report cited above indicates that this service was a recurring reference across the debates, described as the central near-term opportunity for the European Earth observation industry, and as a progressive effort to move from currently fragmented capabilities to a structured governmental service.
Three readings follow for a data provider. The horizon is that of the next budget framework, therefore medium term. The approach is progressive, with studies and a pilot, which opens opportunities before deployment. And the structuring will replace a current fragmentation, which will redistribute positions.
The requirements specific to the field
They define a real barrier to entry.
The conference report states here the point most directly useful to a provider. It reports that defence applications carry specific requirements around data security, secure data flows, military-compliant cloud infrastructure and confidentiality that civilian systems are not necessarily designed to meet, and that the industry needs clarity on this distinction when positioning products and services.
Four distinct requirements read out of that. Data security at rest. Securing the flows. Compliance of the hosting infrastructure. And confidentiality, which covers people as much as systems.
That enumeration matches almost exactly the five layers the medical cluster described, which is no accident: both domains handle data whose disclosure produces lasting harm.
A direct commercial consequence follows. A provider already holding a formalised and certified security arrangement crosses a barrier competitors will have to build, and that lead is measured in months rather than weeks.
Operational uses of Earth observation
Four families differ in their timeliness requirements.
Situational awareness through Earth observation consists of maintaining an updated understanding of an area, through regular observation and change detection. It requires sustained repetition rather than extreme responsiveness.
Damage assessment consists of documenting the state of infrastructure after an event. It requires rapid acquisition and comparison with a prior state, which presupposes an archive.
Maritime monitoring consists of following activity at sea, notably by cross-referencing imagery with position signals. It requires broad coverage and it is particularly well served by radar, whose independence from weather conditions the opening article noted.
And crisis management, whether of natural or human origin, requires acquisition under time constraint and rapid distribution to field actors.
Those four families of use share a common feature. They rest on change detection rather than on describing a state, which returns to the temporal annotation need the constellations article set out.
What sovereignty covers in Earth observation
The term deserves decomposing, failing which it designates nothing precise.
It is used abundantly in Earth observation. It covers in practice four quite distinct requirements better treated separately.
Acquisition sovereignty, meaning the ability to obtain an observation without depending on a foreign supplier who might refuse it.
Transmission sovereignty, meaning the ability to route data through controlled infrastructure, the object of the constellation noted above.
Processing sovereignty, meaning the ability to exploit data without it transiting third-party systems or jurisdictions.
And competence sovereignty, meaning the availability on the territory of the necessary know-how, including that of annotation and data preparation.
That fourth dimension is by far the least discussed and it concerns a data provider directly. A sovereign acquisition capability whose exploitation depended on outside services does not constitute complete sovereignty, and that argument is admissible in a commercial discussion.
Annotation in this context
The requirements described change how the work can be conducted.
Four points show it.
The location of processing becomes a contractual criterion rather than a preference, which the market article flagged for institutional clients.
Clearance of the people working on the data may be required, which narrows the pool and lengthens setup times.
Access traceability becomes a formalised contractual obligation, a requirement whose modalities the medical cluster detailed.
And the nature of the annotated data may itself be classified, which forbids certain working arrangements.
Those four points have an important practical consequence. The field is not accessible through technical competence alone; it presupposes prior organisation, and that organisation is precisely the barrier protecting those who have crossed it.
Dual use and its limits
This notion is used too broadly.
Dual use denotes a technology’s capacity to serve civil and security applications. The institutional analysis cited above notes that the European space strategy for security and defence, published in 2023, acknowledged that satellite navigation, communications and imagery were critical to the military domain and that sovereign solutions strengthening strategic independence through a dual-use approach linking civil and security applications were needed.
Three important limits temper that notion. The mandate constraint noted at the outset, preventing some civilian programmes from serving a defence use whatever their technical capability. Infrastructure and confidentiality requirements, which are not satisfied retroactively on a system designed for open use. And conditions of employment, a defence use presupposing availability and confidentiality guarantees an ordinary commercial service does not provide.
Those three limits largely explain the development of a distinct capability rather than the extension of the existing one, the movement the preceding section described.
The role of commercial suppliers in institutional use
This development changes the structure of supply in Earth observation.
Institutional needs were long served by institutional means. The constellations article showed the commercial offering has grown considerably, and that offering is now being called upon.
Three reasons explain that recourse. Available capacity, a commercial constellation offering a revisit few sovereign means reach. Responsiveness, a commercial operator being able to retask quickly. And cost, buying data often being cheaper than developing an own capability.
Three serious caveats frame it in return. Availability is not guaranteed, a supplier potentially being subject to constraints from its own jurisdiction. The confidentiality of the request itself raises questions, a tasking revealing an interest. And continuity depends on the supplier’s economic health, a point the market article flagged.
Those three caveats explain the mixed configuration being sought: sovereign means for what must be, commercial recourse for densification. That mixed arrangement creates a need to harmonise heterogeneous sources the constellations article described.
The limits the field acknowledges
What Earth observation does not supply in this context deserves setting out.
The field itself is well aware of it. Observation supplies an observation and not an interpretation. An image shows a state; inferring an intention from it is analysis drawing on other sources and it remains human.
It is constrained by acquisition conditions. The opening article showed an announced revisit is not a guarantee of observation, a constraint particularly penalising when the need is time-sensitive.
It produces detections whose false alarm rate conditions usefulness. An arrangement flagging too many unfounded events is abandoned by its users, a mechanism the medical imaging cluster documented in another form.
And it requires validation whose means are not always available, ground verification being precisely what remote observation aims to avoid.
That fourth limit deserves particular emphasis for a data project. Building an evaluation corpus presupposes a reference, and in this domain that reference is often hard to obtain, which makes correctly built and documented corpora all the more valuable.
The less visible ground-level demand
This category escapes the major programmes and it is more accessible.
Security uses are not limited to defence in the strict sense. Four neighbouring domains use the same techniques with lighter requirements.
Civil protection and crisis management, which use imagery to map a disaster area and direct relief. Those actors are territorial public services rather than central administrations.
Critical infrastructure monitoring, energy, transport and water distribution networks, whose operators are often companies.
Civil maritime safety, including pollution response and fisheries monitoring, which belongs to specialised agencies.
And border and flow monitoring, which involves services whose purchasing procedures differ from those of defence.
Those four domains present three advantages for a provider. Security requirements there are real but lighter than in classified defence. Decision cycles are shorter. And they constitute a sector reference usable later with more demanding interlocutors.
That progression is probably the most realistic route into the security domain for an annotation provider not yet established there.
A specific documentary requirement
One technical point deserves flagging because it differs from other uses.
A corpus intended for this domain must document not only its composition, as any serious corpus does, but also the provenance of each of its images and the conditions under which they were obtained.
Three reasons justify that. Licence conditions on commercial sources may restrict use, a question the market article raised and which becomes decisive here. Data traceability is a contractual requirement, presupposing that the origin of each element can be established. And reproducing an evaluation presupposes knowing which data it covered.
One practical consequence follows. A corpus assembled from diverse sources with no provenance register will not serve in this domain, whatever the quality of its annotation.
That constraint is easily satisfied if anticipated from the outset, and it becomes insoluble afterwards, which matches the principle established in earlier clusters: documentation is produced during, never after.
What this domain shares with medical imaging
The two sectors present structural similarities rarely noted.
Four commonalities appear. Supplier evaluation covers the security arrangement before it covers work quality, which makes the security file a decisive commercial document. Access and operation traceability is a formalised contractual requirement rather than good practice. The location of processing is negotiated upstream. And decision cycles are long, procedures formalised and relationships durable once established.
Two differences separate them, however. The legal regime is not the same, health data protection belonging to an explicit framework where defence requirements belong to contracts and clearances. And the certification required differs, a medical device presupposing a regulatory marking with no equivalent here.
That comparison has a strong practical consequence for a provider already established in one of the two domains. Most of the organisational investment, security arrangement, traceability, incident procedure and documentation, is reusable, which makes access to the second domain markedly cheaper than a standing start.
Approaching this Earth observation market
Four steps structure a commercial approach and their order matters.
Establish your own security arrangement first and document it, since the field assesses that point before any other, as the medical cluster showed for a neighbouring sector.
Then identify the right interlocutor. Large defence industrial groups subcontract part of their data needs, and they are often a more accessible entry point than the administrations themselves.
Position on genuinely scarce competences, the temporal annotation and change detection described above, rather than on volume.
And make the competence sovereignty argument set out above, which distinguishes a European provider and which few state explicitly.
One observation usefully completes that approach. Cycles in this market are long, procedures formalised and documentary requirements heavy, but relationships established there are durable and engagements better valued than in ordinary civilian uses.
What the fragmentation means today
One structural feature of the current European Earth observation situation deserves stating, since it shapes near-term opportunity.
Capability today is distributed unevenly. A few member states operate their own remote sensing satellites, others rely on partnerships or commercial purchase, and the collective programme serves civilian purposes. The result is that similar needs are met through dissimilar arrangements across the continent.
Three consequences follow for a provider. Requirements vary by country rather than being harmonised, which means a reference established in one member state does not automatically transfer to another. Procurement happens at national level in most cases, which favours providers with national presence and relationships. And the structuring effort described above will change this, which makes the present period one where positions are still being taken.
That third point is the operative one. A structured service replacing fragmented arrangements will consolidate demand and formalise requirements, and providers already known to the relevant actors when that happens are better placed than those arriving afterwards.
The practical implication is that visibility now matters more than capacity now. Building relationships, references and a documented security posture during the fragmented period costs less than competing for position once demand has consolidated.
Common errors of reading
These misreadings recur often enough that naming them is usually enough to avoid them.
- Assuming civilian data becomes usable in defence by simple decision.
- Treating dual use as a technical property rather than a regime of employment.
- Approaching this market without a formalised and documented security arrangement.
- Using the term sovereignty without distinguishing its four dimensions.
- Neglecting competence sovereignty in a commercial argument.
- Addressing administrations directly rather than the industrial groups that subcontract.
- Offering volume where the scarce competence is temporal.
- Underestimating the time needed to put required clearances in place.
- Ignoring that the location of processing is a contractual criterion here.
- Assuming a system designed for open use retroactively satisfies these requirements.
- Neglecting civil security uses, more accessible than classified defence.
- Inferring an intention from an observation without drawing on other sources.
Why the barrier is organisational rather than technical
A closing observation clarifies what actually separates providers in this domain.
The annotation work itself is not harder here than elsewhere. Detecting a vessel, delineating an installation or comparing two dates uses the same techniques the civilian articles in this series describe, and a team competent in one is competent in the other.
What separates providers is everything around the work. A documented security arrangement. A traceability mechanism that functions. Cleared personnel where required. A provenance register for every image used. An incident procedure that exists before an incident. And documentation produced during the work rather than reconstructed afterwards.
None of those is technically difficult. All of them take time to establish, and none can be improvised when a client asks.
That is why the barrier is durable. A competitor can match annotation quality in a quarter; matching an organisational posture built over two years takes two years. For a provider who has already made that investment for another regulated domain, the same posture applies here with modest adaptation, which is the most valuable thing this article has to say to anyone in that position.
What to take away
Space-based Earth observation is described as having moved from a primarily civilian and environmental capability to a core component of European security and defence, which changes technical requirements and market access conditions.
Three readings emerge. The separation between civil and defence is real and legal, a civilian mandate being able to prevent a programme’s use in a defence context whatever its technical capability, which led to developing a distinct governmental capability rather than extending the existing one. Defence applications impose data security, secure flow, compliant infrastructure and confidentiality requirements that civilian systems are not designed to meet, and those requirements constitute a barrier to entry few providers have crossed. And sovereignty covers four distinct dimensions, acquisition, transmission, processing and competence, the last being the least discussed and the most directly relevant to a European data provider.
For a use with different economic requirements, the article on Earth observation and insurance examines risk assessment. For the European position as a whole, the article on European sovereignty and Earth observation covers the strategic question.
To explore delivery arrangements, supported formats and applicable control mechanisms, see our dedicated page on geospatial data processing. And if you need an annotation provider meeting formalised security requirements, let us discuss your project.
