Earth Observation has spent years getting better at detecting hazards. For many disaster-related applications, the commercialisation problem now sits further downstream, where observation has to enter an institutional decision and remain there long enough to become routine.
California as a Commercialisation Test
When I was flying regularly to California for sales meetings at Planet, wildfire monitoring was already one of the native EO use cases. MODIS active-fire products had already been available for years. Fires cover large areas, develop quickly, can be difficult to access from the ground and have severe consequences. Better visibility had immediate operational value.
Nearly a decade later, California can observe fire in far more ways: optical imagery, SAR, thermal sensors, dedicated wildfire constellations, aircraft, camera networks, AI detection, predictive models and multi-source platforms now contribute to wildfire intelligence.
The 2025 Los Angeles fires happened inside that much denser observation environment. Their severity says little by itself about the performance of EO. Wildfire outcomes depend on weather, fuels, land use, ignition, emergency management and many other factors. What the California timeline does show is how rapidly observation capability has expanded while the surrounding institutional system has had to evolve with it.
California is useful here because it removes the easy explanation that this is mainly a European public-procurement issue.
In 2019, the state created an Innovation Procurement Sprint for wildfire technology. Technosylva received a $383,000 proof-of-concept contract for real-time fire-spread prediction. By the following year, CAL FIRE had moved to statewide implementation. The 2020–21 state budget included $4.4 million for implementation and $7.6 million in ongoing funding for the capability. FIRIS and ALERTCalifornia followed the same path into operational use.
The same system also filters products out. PG&E tested automated wildfire-detection camera vendors and did not select the first group for full operational deployment because their performance did not meet operational criteria. A later vendor was contracted under refined requirements.
These examples make procurement only part of the story. Technical performance in the operating environment matters. So do policy, communications, staffing, training, system integration, authority and the ability of an institution to maintain the capability after a pilot ends.
The Los Angeles after-action work makes that surrounding system visible. Its findings covered outdated policies, inconsistent practices, communications vulnerabilities, multi-agency coordination and weaknesses in operational systems. Extreme winds grounded aircraft during a critical period and removed one source of real-time situational awareness. Subsequent actions include additional data feeds into command systems, satellite-enabled communications, dispatch upgrades, more multi-agency training and additional satellite hotspot monitoring.
A sensing product entering disaster management therefore lands inside an existing operational system, with its own constraints and responsibilities.
The Budget Behind Preparedness
On 1 October, at CASSINI in Prague, I was reminded how much disaster and climate risk still features in the space startup pipeline. The same day, ESA Space Solutions published an article promoting its current Wildfire Preparedness & Risk Assessment opportunity.
The underlying ESA call opened in September and funds proof-of-concept studies and pilot projects. Its focus has moved upstream into preparedness: risk assessment, fuel-load management, prevention and adaptation. Applicants are expected to show customer engagement, identifiable customer needs and a path to a commercially viable service, alongside technical feasibility.
There is a useful economic detail in ESA’s framing. Citing UNEP, it notes that more than half of wildfire-related expenditure commonly goes to response, while planning typically receives only 0.2% of the total wildfire budget.
That distribution matters for commercialisation. A company can produce valuable information for pre-season planning and still be selling into a function with a very different budget structure from emergency response.
Technical value and addressable expenditure do not develop at the same pace.
The new ESA call is not evidence that wildfire EO has been trapped in pilots. Proof-of-concept work is a legitimate part of bringing new capabilities into use, and ESA explicitly requires commercial and customer evidence. The more interesting long-term observation is that the innovation pipeline continues to replenish itself. New sensors, analytics and services keep arriving, while each of them eventually has to find a stable place inside an institution.
Where the Market Sits
That place becomes easier to see once “wildfire” stops being treated as a market in itself.
A company can sit in a large and obvious use case without having found a repeatable market.
Wildfire intelligence can support pre-season planning, dispatch, resource allocation, evacuation, utility operations, infrastructure protection and insurance. Those activities have different users, budget owners, decision cycles and procurement routes. The technical capability may be similar while the commercial path changes substantially.
Flood monitoring shows the same pattern. The European Flood Awareness System has been operational for years and provides probabilistic flood forecasts across Europe. EFAS supplies information to national and regional authorities, but it does not control the actions they take. Authorities combine those forecasts with national systems, procedures and local information. The forecast can be mature while the operational decision belongs somewhere else.
That division of responsibility can also obscure where the commercial market sits. EFAS is part of the Copernicus Emergency Management Service, which is free to authorised users. Free to the user, however, does not mean that there is no commercial market behind the service. CEMS Risk and Recovery activations are translated by the Joint Research Centre into technical specifications and tendered to service providers. A 2026 JRC procurement for the CEMS Hydrological Forecast Centre – covering daily analysis and dissemination of EFAS and GloFAS forecasts – was estimated at €4.55 million. The free public layer can therefore sit on top of a paid B2G supply chain.
Insurance is particularly useful as a comparison because this is already happening at portfolio scale. Across ESA BASS, nearly three-quarters of Finance & Insurance Demonstration Projects have generated sales. The 63 projects collectively reported €17.71 million in sales, with a quarter secured outside Europe.
Insurance provides a more tightly defined route from information to consequence. In Laos, satellite and ground data have been used inside sovereign disaster-risk insurance arrangements with coverage and payout rules defined in advance. In 2025, Laos paid a $3.4 million premium for two years of coverage worth up to $16 million. After severe flooding affected more than 300,000 people, a $2 million payout was made within six business days.
Here the data enters a financial mechanism that already has a buyer, contract, threshold and defined consequence. The commercial structure is visible because the institutional decision is visible.
Infrastructure provides another version. Deutsche Bahn uses satellite-derived vegetation information to support inspection and vegetation-management teams along the rail network. The information helps prioritise field work inside an existing maintenance process with an operational owner and an established reason to spend money.
From Pilot to Recurring Adoption
Dr Alexander Barinov, BAA International’s Founder and CEO, approaches the same question from the perspective of Geotechnical Engineering & Infrastructure Resilience in Building Resilient Infrastructure in Developing Regions. He describes monitoring programmes that continue collecting evidence without a defined end point or an obligation to act. Data can accumulate for years while the underlying infrastructure decision is postponed. His proposed structure is time-bounded monitoring that closes with formal risk-reduction recommendations delivered to the institution responsible for the next step.
For an EO company, the same issue appears commercially when a technically useful product remains adjacent to the decision process rather than becoming part of it.
A successful pilot can establish technical usefulness, expose a product to real operating conditions and show that a particular team can work with it. Recurring adoption asks more of both sides: continued product performance, budget continuity, responsibility, integration, training, procurement and enough organisational stability for the capability to become routine. The distance between technical maturity and recurring adoption is where many EO companies spend years.
Some institutions already have a recurring function into which EO can fit. Maintenance budgets exist because infrastructure has to be inspected and maintained. Insurance premiums exist because risk is priced and transferred. Disaster-risk finance can define thresholds and payouts before an event occurs. These structures give information somewhere operational and financial to go.
Preparedness can be harder. The need is clear, but the budget architecture may still favour response. The ESA/UNEP figure makes that mismatch unusually visible. Better risk intelligence can improve where resources are positioned before a fire season, yet the planning function receiving that intelligence may control only a small share of total wildfire expenditure.
For disaster-related EO companies, market analysis therefore needs to go below the level of the hazard or use case. The useful evidence is the decision affected by the observation, the institution responsible for that decision, the authority attached to the role, the workflow in which the information appears, the budget supporting the function and the route by which the capability can remain in use.
This is a more demanding commercial test than identifying a large problem and demonstrating that satellite data can help. It also explains why technically mature disaster-monitoring capabilities can follow very different commercial trajectories.
Institutional support is also moving further downstream. ESA BASS now works with insurers, reinsurers and specialist gateways to bring market needs, champion users, pilot opportunities and routes to market into the support structure itself.
Three Observations
Treat institutional capacity as part of product-market fit. A technically useful service still has to be absorbed by the organisation that will use it. California shows both outcomes: capabilities that moved into statewide operations and technologies that did not survive operational testing. Product-market fit in disaster EO includes the institution’s ability to integrate, operate and sustain the product.
Follow the budget inside the use case. A large and urgent problem does not imply a large accessible budget. Wildfire preparedness makes this unusually visible: more than half of wildfire-related expenditure commonly goes to response, while planning receives only 0.2%. For an EO company, the accessible pool is narrower again: satellite-derived intelligence competes for a share of the planning and preparedness budget rather than the entire wildfire budget. The relevant market is the budget attached to the decision the product supports.
Look for an existing function that can make the need recur. The strongest examples here sit inside maintenance, insurance and disaster-risk finance. The observation enters something an institution is already responsible for doing, with an operational owner and a mechanism that can continue paying for the capability.
A hazard does not create a market. A recurring institutional decision does.