A farmer wants to know which field to inspect. An insurer is looking to identify buildings affected by flooding. A railway infrastructure manager needs to prioritize embankments for monitoring. None instinctively asks for a satellite image: all expect an actionable answer. It is in this gap between observation and action that a space business can emerge, without a factory, a rocket or a satellite. Looking ahead to September 2026, this entrepreneurial avenue deserves attention. The prospects outlined here build on developments already documented; they do not assume that every application has reached maturity.
A satellite is no longer a prerequisite
For a long time, starting a space business primarily meant substantial industrial investment. Earth observation has opened another door. Europe’s Copernicus program makes Sentinel data freely available, while Landsat provides a public archive spanning several decades. Commercial providers, meanwhile, offer higher-resolution imagery or complementary capabilities. A startup can therefore begin by buying computing resources and expertise rather than orbital hardware.
This accessibility does not make the data immediately usable. Images need to be corrected, cloud cover handled, acquisitions compared and observations linked to what is happening on the ground. Optical imagery provides information about vegetation and land cover, among other things; radar enables certain observations at night and through clouds. But no sensor answers every question. The product to sell is not a view from space: it is a better-informed decision.
Agriculture: sell a useful inspection round, not a colorful map
In agriculture, one initial application is detecting differences in growth within a field. Image time series can highlight an area where vegetation is lagging. The service becomes tangible when it tells a field technician where to investigate, with historical context and a priority level. The entrepreneur does not necessarily replace the agronomist: instead, the service saves them from having to inspect every field with the same urgency.
The pitfall would be automatically turning a signal into a diagnosis. A decline in a vegetation index may reflect a lack of water, disease, harvesting or a soil problem. Recommending an action requires combining weather data, crop calendars, soil information and field observations. An irrigation recommendation requires more than an attractive image: it also depends on available equipment, restrictions and the farm’s objectives.
The most accessible market is not always direct sales to individual farmers. Cooperatives, agricultural merchants and consultancies already have clients and field technicians. They can integrate an alert into their existing tools. Looking ahead to 2026, the strongest commercial proposition therefore remains a specialized service tailored to a particular crop and region, rather than a platform claiming to optimize all of global agriculture.
Insurance: speed up assessments without automating unfairness
After a disaster, satellite observation can help map a flooded or burned area and match it against a portfolio of insured properties. The operational benefit is immediate: estimating exposure, organizing site visits and assigning teams. Access to a relevant image nevertheless depends on the satellite’s pass, the sensor, acquisition conditions and processing times. Any commercial promise must account for these limits, not conceal them.
Parametric insurance offers another avenue. Payouts then depend on an indicator defined in the policy, for example one linked to rainfall or vegetation conditions, rather than a conventional individual loss assessment. But this approach carries basis risk: the indicator may not accurately reflect the actual loss. A credible business will document failures and uncertainties rather than present its model as an infallible arbiter.
To enter this market, it is better to address one specific step: preparing case files, mapping exposure or helping to triage claims. Making compensation decisions alone entails much greater responsibilities. Insurers will expect reproducible methods, a validation track record and traceable results. The value comes as much from the supporting evidence as from the algorithm.
Infrastructure: identify where to send specialists
Roads, railways, levees and pipelines form networks too extensive to inspect everywhere at all times. Satellite radar interferometry can measure certain ground movements using successive acquisitions. This technique, already in use before 2026, can notably identify subsidence trends. It does not, however, detect every deformation: measurement geometry, vegetation and the stability of the observed surfaces influence the results.
A useful offering turns these measurements into a list of locations to examine, linked to the asset register and past interventions. It distinguishes a persistent signal from an uncertain anomaly. Above all, it does not confuse ground movement with imminent danger. Human inspection and local instrumentation remain essential to understanding the causes and deciding on an intervention. Here, the satellite serves as a large-scale screening tool.
Build a business, not just a demonstration
The first investment should be in understanding the industry. Who receives the alert? In which software? What action can they take? How much does a false alarm or a missed problem cost? These questions determine the product, the data required and an acceptable price. An impressive dashboard that is not part of everyday procedures risks remaining a pilot with no future.
Before developing a complete offering, three things must be validated:
- A recurring problem: find a customer who frequently encounters the same difficulty and has a budget to resolve it.
- Ground-based evidence: compare results with independent observations across multiple situations, without cherry-picking only the successes.
- Viable economics: account for commercial imagery, computing, storage, quality control, support and customer-specific customization.
Public data lowers the barrier to entry, but not necessarily the total cost of the service. Resolution or speed requirements may necessitate additional purchases. Commercial licenses must also permit the intended uses. Sales cycles, meanwhile, can be longer when customers manage critical assets or compensation payments. A pilot should therefore establish its success criteria and the conditions for a paid contract from the outset.
The real competitive advantage is built on the ground
Advances in artificial intelligence should make extracting information easier and reduce some manual tasks. That prospect does not guarantee a profitable business. If several players have access to the same images and models, differentiation will come from elsewhere: lawfully obtained ground data, industry expertise, software integration and user trust. Sensitive information derived from combining imagery with customer records will also need to be protected.
What next? For an entrepreneur launching in September 2026, the prudent path would be to choose a region, an industry and a measurable decision, then demonstrate the service’s usefulness with an initial partner. The prospects are real, but realizing them will depend less on the number of images analyzed than on the number of relevant actions triggered. The next useful space business could look like an industry-specific software company — with a window onto Earth behind the scenes.