The Geospatial Data Stack

What is the Geospatial Data Tech Stack?

Geospatial data is any data that is connected to a place on earth. We often think about satellite imagery, but geospatial data is broader than that. It can be a GPS coordinate, weather related data, a drone image etc.

The landscape I created is about the ecosystem of companies building around this data. From collecting to processing geospatial data in order to turn it into useful decisions.

What are the major types of geospatial data?

Concretely, there are four main forms of geospatial data:

• Maps and location data. Basic geographic information about the physical world, from roads and buildings to land ownership, infrastructure, or “administrative” boundaries.

• Satellite and aerial imagery. Images of the Earth collected from above. Whether by satellites, aircraft, drones, or other aerial systems. Depending on the use case, these images can capture visible light, heat, radar signals, or other forms of sensing.

• Signals and sensor data. Location based data that does not necessarily look like an image. It can come from ships, vehicles, weather systems, connected devices, radio signals, or atmospheric measurements.

• Derived geospatial intelligence. Raw geospatial data is transformed into something easier to act on, such as a risk score, an alert, a prediction, or a recommendation.

What does the geospatial data tech stack look like?

I split my map into a three layer stack:

• Layer #1: Proprietary geospatial data sources. These are the companies creating or unlocking proprietary geospatial data. This includes satellite imagery, SAR, hyperspectral imagery, RF signals, weather data, maritime data, drone data, street level data, mobility data, and sensor networks. The important thing to keep in mind is that often these companies also provide APIs, analytics, or dashboards in addition to raw “geospatial data”. But their main asset is the data source itself.

• Layer #2: Geospatial data infrastructure. These are the tools that make geospatial data easier to access and use. They usually do not collect proprietary raw data themselves. Instead, they help customers turn fragmented geospatial data from many sources into something usable. This means making the data easier to find (marketplaces of data providers), access, clean, process, combine, and integrate into the customer’s workflows.

• Layer #3: Geospatial intelligence & applications. These are the companies turning geospatial data into decisions. This is basically the layer where geospatial becomes a verticalization play. The same geospatial data can create very different products depending on the industry: From crop monitoring for agriculture to risk analysis for insurance, movement visibility for logistics, or intelligence for defense. Here, the customer is not really buying geospatial data. They are buying an answer to a business question.

Is this an emerging or mature market?

The geospatial data stack is not new. Governments have long used maps to manage land and public infrastructure. Militaries use satellite imagery to monitor other countries. Utilities use GIS to monitor power lines, pipes. Insurers use location data to assess exposure to floods, fires, or storms. etc.

This market has already gone through several major evolutions:

• From paper maps to digital GIS, roughly 1960s to 1980s. The first big shift was the digitization of geographic information. Instead of working with static maps and manual surveys, organizations could store, layer, query, and analyze geographic data inside specialized GIS software. This made geospatial data a professional software category, mostly used by experts.

• From government satellites to commercial Earth observation, roughly 1970s to 2000s. Satellite imagery was initially driven by governments, defense, and scientific programs. Over time, commercial Earth observation companies made this data more accessible to private companies. The result was a larger market for imagery. Not only for states and researchers, but also for agriculture, insurance, energy, logistics, and finance.

• The consumerization of location data, roughly 1990s to 2010s. With GPS, smartphones, navigation apps, and consumer mapping products, location data became part of daily life. Maps were no longer only specialist tools. They became consumer interfaces.

• The democratization of geospatial data,  roughly 2010s to today. Historically, geospatial workflows often meant large local files and formats that were difficult to work with and access. Over the past decade, the stack has moved into the cloud and become much easier for software teams to use. This makes geospatial data easier to integrate into modern software products.

Why is this ecosystem changing now?

Now, mostly thanks to AI, the geospatial data stack is going through a new evolution.

• There is more data, but also too much data to interpret manually. The world is being measured more often and with better precision. There are more satellites, more drones, more sensors, more public datasets, more commercial datasets, and more location based signals than before. But this creates a new problem: collecting more imagery or location data is not enough if users cannot extract the right signal from it.

• AI is turning geospatial data into something more readable. Historically, a lot of geospatial analysis required specialists or narrow models built for one specific task. AI makes it easier to detect what changed, identify what matters, and transform raw observations into usable signals. Basically expanding the scope of what we can do with geospatial data.

• Intelligence is moving closer to the point of collection. The interesting shift is not only that AI can analyze satellite images after they are downloaded. Over time, satellites, drones, and sensor networks can start filtering what they see directly. Instead of sending everything back, they can prioritize what matters first (and this is pretty amazing btw).

• Geospatial workflows are becoming easier to put in place. Users don’t need GIS experts to work with geospatial data. They can increasingly ask questions, define constraints, and get answers through more intuitive software interfaces.

Geospatial Data & Access

What is this category about?

• This category covers companies that create new data about what is happening in the physical world and make that data available to other systems. Satellite imagery is the most established example, but the category is much broader. A geospatial signal can describe what a place looks like, what is happening in the atmosphere above it or what is moving through an area. The common idea is that the company owns or controls part of the sensing layer rather than simply analysing data collected by somebody else.

• The first problem is that much of the physical world is still observed too infrequently. A satellite may pass over an area only at certain times, while ground measurements are often concentrated around populated regions. This creates gaps when customers need to know what is happening now rather than what happened during the last available observation. Many companies in the category therefore compete on revisit frequency or continuous collection rather than resolution alone.

• Another problem is that normal imagery can only reveal part of what is happening. Clouds can prevent optical satellites from seeing the ground, while many economically useful signals are invisible to a normal camera in the first place. This creates room for sensing systems that measure different physical properties and allow customers to observe something that was previously difficult to detect remotely.

• Some environments remain poorly measured because deploying traditional sensors is expensive. The atmosphere above oceans is one example, while large agricultural regions can also lack dense ground observations. Instead of waiting for public sensing infrastructure to improve, startups are increasingly building their own distributed networks and turning the resulting observations into commercial datasets.

• The underlying need is gradually shifting from maps toward machine readable observations. Software increasingly wants to know whether something changed rather than display an image to a human analyst. Better geospatial data therefore becomes an input into forecasting systems and automated decisions. The more frequently those systems operate, the more valuable fresh observations become.

What do products in this category do?

• The largest group collects Earth observation data from satellites. The data varies significantly depending on the sensor. Planet and BlackSky collect optical imagery that shows visible changes on the ground over time. ICEYE and Umbra collect radar data, which can observe the surface at night and through cloud cover. Pixxel and Wyvern collect hyperspectral data that captures how materials reflect light beyond what a normal image can reveal. The important distinction is that these companies are not simply producing different resolutions of the same picture. They are measuring different properties of the physical world.

• Atmospheric sensors create weather data in places where traditional observation networks are sparse. WindBorne and Sorcerer send autonomous sensors through the atmosphere using balloons. Meteomatics uses weather drones to collect measurements at different altitudes. Skyfora takes a different approach by extracting atmospheric information from GNSS signals received by telecom infrastructure. These networks create observations that can be fed into weather models, especially in regions where existing measurements are limited.

• Ocean sensors create persistent data about conditions that are difficult to observe remotely. Sofar Ocean deploys connected buoys that measure local ocean conditions and transmit the observations continuously. Andrenam uses distributed sonar buoys to detect underwater activity and turn acoustic signals into location based maritime awareness. In both cases, the sensing network produces information that satellites alone cannot reliably provide.

• Agricultural sensors can map biological activity rather than the physical appearance of a field. BioScout places autonomous sensors directly in farming regions to capture airborne spores and identify crop pathogens. Each observation is tied to a location, allowing growers to see how disease pressure develops across an area. This creates a geospatial dataset from something that would otherwise require manual sampling and laboratory analysis.

• Some geospatial datasets are built from signals rather than environmental measurements or imagery. HawkEye 360 detects radio frequency emissions from satellites and geolocates where they originate. This allows customers to map activity that may have no visible signature at all. The underlying principle is important for the category: geospatial data does not have to describe what a place looks like. It can describe what is being emitted or happening at that location.

What types of customers are using these products?

• Defense and government agencies are among the strongest buyers because persistent observation has direct operational value. HawkEye 360 has integrated its RF intelligence into military exercises alongside systems from Lockheed Martin. World View was selected as the high-altitude balloon provider for UNITAS 2025, where its sensing platform supported a multinational naval exercise. These customers care much more about how quickly an observation can affect a decision than about receiving another dataset.

• Weather sensitive enterprises buy geospatial data when forecast error has a direct operating cost. Tomorrow.io deployed its aviation weather platform across Lufthansa's network. Meteomatics has worked with Swissgrid to centralize weather information used in grid operations. In both cases the data sits close to an operational workflow where better information can change what the customer does during the day.

• Insurance is becoming a meaningful customer group because geospatial observations can help establish what actually happened at a location. MAIF uses Meteomatics' MetX Claims product to verify localized weather events during claims handling. Covéa uses the platform to analyse weather risk and anticipate claims activity. This is a good example of geospatial data becoming part of a business process rather than remaining a specialist mapping tool.

• Maritime operators use private sensing networks when public ocean data is not precise enough for operational decisions. MOL Group deployed Sofar Ocean's Wayfinder after trials showed fuel savings from better routing. Dorian LPG later reported similar savings across its fleet. The value comes from combining better ocean observations with decisions that already carry significant fuel cost.

• Agriculture provides another route to commercialization when new observations can change when growers intervene. BioScout's GRDC backed network has deployed sensors across Australian farming regions so growers can see disease pressure before spraying. Its customer cases in New Zealand show that the data can reduce unnecessary fungicide applications, giving the sensing layer a relatively direct economic value.

What are the major trends shaping this category?

• Governments are moving from experimenting with commercial geospatial data toward buying it as part of their permanent infrastructure. The NRO's commercial imagery contracts with providers such as Planet and BlackSky are worth billions of dollars over their duration. In 2026 it expanded commercial procurement further into new sensing modalities. NOAA is moving in the same direction and launched a new commercial environmental-data acquisition vehicle with an estimated ceiling of $2.35 billion over ten years. Commercial sensing is increasingly becoming part of the public observation architecture rather than sitting beside it as an experiment.

• The market is moving beyond optical imagery toward combining different ways of observing the same place. In February 2026 the NRO explicitly described its commercial remote sensing strategy as multi-phenomenology and included radar alongside newer sensing approaches. A later round of awards included ICEYE and Pixxel. This matters because the future geospatial stack is unlikely to have one dominant data source. Different observations answer different questions, which expands the market for companies creating genuinely new signals.

• AI is making the observation layer more important rather than removing the need for it. ECMWF's operational AI forecasting system can generate forecasts much more efficiently than traditional numerical models, but it still starts from an estimate of current conditions built from around 60 million quality controlled observations. ECMWF explicitly notes that machine learning forecasting still depends on data assimilation to establish the starting state. As forecast models become easier to build and run, differentiated real-time observations may become one of the harder assets to reproduce.

• Simply putting satellites in orbit is becoming less scarce. ESA estimated around 16,000 functioning satellites were in Earth orbit at the end of July 2026. The number of commercial constellations continues to increase, particularly in low Earth orbit. This expands the amount of data that can be collected but also changes the competitive equation: owning a satellite is becoming less unusual, so companies increasingly need to differentiate through what they measure or how often they can measure it.

How does the funding environment look?

• This remains an unusually capital intensive data category, and the largest companies are raising amounts closer to infrastructure financing than conventional software rounds. ICEYE's 2026 financing transaction reached €1 billion, including €450 million of new primary capital. Tomorrow.io raised $175 million in February 2026 and added another $35 million a few months later. Investors are effectively financing the construction of proprietary observation networks before those networks can generate software-like data revenue.

• Public markets have become part of the financing path for companies that reach sufficient scale. HawkEye 360 raised $416 million in its May 2026 IPO at $26 per share. Several earlier companies in the category had already reached public markets, showing that Earth-observation businesses do not necessarily need to be sold to an incumbent once their infrastructure reaches scale. The trade-off is that investors can now see much more clearly how expensive these businesses remain to operate.

• There is also meaningful funding flowing toward completely new sensing networks rather than only additional satellite constellations. WindBorne raised a $37 million Series B in August 2026 to expand its balloon network and weather platform. Andrenam raised an $18 million Series A for its distributed underwater sensing system. That is interesting because the investment thesis is not simply that the world needs more imagery. Investors are financing new ways of measuring areas where useful data is still scarce.

• Government procurement can act almost like a second financing system for the category. WindBorne has received U.S. Air Force funding tied to its weather data work, while Tomorrow.io moved from NOAA validation into a paid commercial data contract. These contracts can help finance the sensing infrastructure while simultaneously proving that the data is useful to a demanding customer. That makes government adoption unusually important to company formation in this market.

What is the defensibility of these startups?

• The strongest moat appears when the company builds a dataset that becomes more valuable with time. Planet is difficult to replicate not simply because it owns satellites, but because it has accumulated years of frequent observations of the same places. A new competitor can launch a similar camera, but it cannot recreate yesterday's archive. The same logic applies to networks where continuous measurements gradually create a unique historical record.

• Unique sensing modalities can create real defensibility because competitors cannot substitute the data with a better algorithm. Radar observations from ICEYE answer questions optical imagery cannot reliably answer. HawkEye 360 captures RF activity that imagery does not contain at all. A model can become better at interpreting a dataset, but it cannot infer information that was never observed in the first place. This is where I see the strongest underlying data moat.

• Owning satellites by itself looks increasingly less defensible. More startups can launch constellations, while space infrastructure is making it easier to get new sensors into orbit. If two providers eventually offer similar resolution with similar revisit frequency, the data starts looking interchangeable and customers gain pricing power. The most vulnerable businesses are therefore the ones whose main differentiation is simply that they have their own satellite.

• The most interesting defensibility may come from closing the loop between proprietary observations and the product built on top of them. WindBorne's balloon network feeds its own weather models, while Tomorrow.io is building sensing infrastructure specifically around its forecasting platform. Better observations can improve the product, which increases usage and creates more reason to expand the sensing network. But this moat only holds if the proprietary data produces meaningfully better outcomes. If open data or competing networks reach similar quality, the expensive sensing infrastructure can quickly become a liability rather than an advantage.

Geospatial Data Infrastructure

What is this category about?

• This category covers the infrastructure that sits between geospatial data and the people or software that need to use it. The previous category is about creating observations of the physical world. This one is about making those observations easier to find and turn into something that can be consumed inside another product or workflow.

• The first problem addressed is fragmentation. A company that needs satellite imagery may have to deal with different providers that each use their own ordering process and delivery format. This becomes especially difficult when a customer wants to combine several sources. Infrastructure platforms try to hide that complexity behind a common interface.

• The second pain point is that geospatial data remains unusually difficult to work with at scale. Earth observation files can be extremely large, while spatial queries require different infrastructure from conventional business data. Moving files between systems also becomes expensive. A growing part of the category therefore focuses on making geospatial data behave more like modern cloud data.

• There is also a latency problem before satellite data even reaches the ground. A satellite can collect far more information than it can immediately transmit. Sending an entire image to Earth before deciding whether anything useful is inside creates unnecessary delay. This has created a new infrastructure layer where processing happens directly onboard the satellite.

What do products in this category do?

• A first product type aggregates geospatial data behind a common access layer. The underlying problem is that buyers do not want to integrate separately with every satellite operator they use. SkyWatch lets customers access imagery from many providers through the same platform and has gradually expanded from satellite data into aerial imagery. UP42 follows a similar model, combining discovery with ordering while normalizing how different datasets are delivered. These businesses are effectively building the procurement layer for geospatial data.

• Some platforms provide infrastructure to the data provider rather than the buyer. Satellite operators need their own systems for commercializing what their sensors collect, but building ordering and delivery infrastructure is not necessarily where they want to differentiate. Arlula's Sales Engine lets providers expose their data without building the full commercial stack themselves. Spiral Blue selected it as the distribution layer for its planned LiDAR constellation. The model is closer to commerce infrastructure for geospatial data than to a conventional marketplace.

• Another group makes large geospatial datasets easier to store and compute against in the cloud. The logic is to avoid repeatedly downloading files into specialist GIS systems before analysis can begin. Wherobots brings large scale spatial processing into a cloud data environment and increasingly exposes it to AI workflows. Earthmover is focused on multidimensional scientific data such as weather datasets, while Ellipsis Drive provides infrastructure for operating directly on large spatial datasets. These products move computation closer to where the data lives.

• In-orbit infrastructure tries to process information before it is downlinked. The reason is simple: transmitting raw satellite data can be slower than analysing it. Ubotica runs AI onboard satellites so useful observations can be identified immediately. SkyServe provides software for deploying models onto satellite compute, while Zaitra combines onboard processing with compression. EDGX goes further into the hardware layer with a space qualified computer running its own software stack. The satellite effectively starts behaving like an edge computing node.

• Loft Orbital lets companies put technology in space without building an entire satellite themselves. A customer can bring the sensor or software it wants to fly, while Loft provides the satellite platform and handles much of the mission infrastructure. Its newer AI products extend the same idea to computing in orbit. The broader bet is that launching a new geospatial product should not require every company to build its own spacecraft from scratch.

What types of customers are using these products?

• Satellite operators are an important customer because many want to focus on their sensor rather than build every layer around it. Spiral Blue selected Arlula to handle distribution for its planned LiDAR constellation. EarthDaily uses Loft Orbital infrastructure to deploy its Earth observation constellation. These companies are effectively outsourcing infrastructure that does not directly differentiate the data they sell.

• Software companies use these platforms when they want geospatial capabilities without becoming geospatial infrastructure companies themselves. Scribble Maps integrated Earth observation purchasing through SkyWatch BUILD. This allows geospatial functionality to appear inside another application while SkyWatch handles much of the complexity behind the scenes.

• Government and scientific organizations are important buyers of the more technical infrastructure. NASA JPL is working with Loft Orbital to fly AI software on its infrastructure. It is also involved in programs with Ubotica and SkyServe around autonomous onboard processing. Arlula has worked with NTT DATA on infrastructure supporting a Japanese government satellite mission. These buyers often become early adopters because they are willing to test infrastructure before it becomes a standardized commercial product.

• Defense demand increasingly appears when geospatial infrastructure becomes real-time infrastructure. Helsing partnered with Loft Orbital on an AI enabled satellite constellation for European defense applications. Fugro has worked with Ubotica on maritime intelligence infrastructure. The requirement here is not simply access to imagery. Buyers want the infrastructure to shorten the path between observation and action.

What are the major trends shaping this category?

• Geospatial data is becoming much more standardized at the infrastructure level. STAC became an OGC Community Standard in 2025 and now supports far more than satellite imagery. The important effect is that providers can expose very different geospatial assets through a common metadata structure. This lowers the amount of custom integration required every time a new dataset enters a workflow.

• Geospatial infrastructure is moving into the same cloud environments companies already use for the rest of their data. Microsoft made Planetary Computer Pro generally available in June 2026 with support for modern geospatial data structures such as Zarr. Databricks now exposes H3 spatial operations directly inside its platform. Geospatial analysis is therefore becoming less isolated from conventional enterprise data infrastructure, which creates pressure on specialist platforms to integrate deeply rather than remain standalone GIS environments.

• Onboard AI is moving from demonstration missions toward actual infrastructure. NASA announced in May 2026 that the Prithvi geospatial foundation model had been successfully deployed in orbit. JPL's FAME program is testing autonomous processing across seven spacecraft in 2026 and plans to expand the system substantially. ESA's Φsat-2 has already demonstrated onboard processing in operational science mode. This strengthens the case that part of the geospatial compute stack will eventually move from ground data centers into orbit.

• The infrastructure is increasingly being designed for machines rather than human GIS users. OGC explicitly framed its 2026 infrastructure work around a world where systems communicate directly and where standards must therefore be machine readable. This is an important shift for the category. Geospatial infrastructure historically helped analysts find data. The next generation also needs to let software discover what data exists and invoke the appropriate processing automatically.

How does the funding environment look?

• The amount of capital required depends heavily on how close the company gets to physical space infrastructure. Loft Orbital raised a $170 million Series C in 2025 after a previous $140 million Series B. That scale makes sense because it is financing reusable satellite infrastructure and mission capacity. It looks much more like an infrastructure business than a conventional geospatial software company.

• Onboard computing sits between those two financing profiles. Ubotica raised an $11 million Series A in 2026, while EDGX raised €2.3 million in 2025. These companies still need to validate technology in orbit, which makes them more capital intensive than pure cloud software. Public space programs can absorb some of that validation cost, reducing how much must come directly from venture investors.

• Strategic acquisition is another credible outcome for the aggregation layer. Neo Space Group completed its acquisition of UP42 in July 2025. That makes sense structurally: a marketplace becomes more strategically valuable when combined with a company that owns data or space assets. I would expect further consolidation where large geospatial groups acquire the infrastructure through which customers already access third-party data.

What is the defensibility of these startups?

• Pure data aggregation is useful but does not create a particularly strong moat by itself. If the underlying satellite imagery belongs to somebody else, another platform can theoretically integrate the same suppliers. SkyWatch and UP42 therefore become more defensible as they move beyond catalog aggregation and embed procurement directly inside customer workflows. The harder they are to remove without disrupting how a customer operates, the stronger the business becomes.

• Cloud infrastructure can become highly defensible through technical depth and data gravity, but it faces powerful horizontal competitors. Wherobots or Earthmover can become difficult to replace once large datasets and production workloads depend on them. The risk is that Databricks, Microsoft or the major cloud providers continue absorbing geospatial capabilities into their general platforms. Specialist infrastructure therefore needs to remain meaningfully better at geospatial workloads rather than merely offering features the horizontal platforms eventually copy.

• In-orbit infrastructure has stronger technical barriers, but standardization could eventually weaken them. Getting software or compute hardware reliably deployed in space requires flight heritage that a new competitor cannot reproduce quickly. Ubotica, Zaitra and EDGX gain credibility every time their systems work in orbit. However, if satellite platforms eventually expose standardized onboard compute environments, part of today's proprietary infrastructure could become a commodity layer. The strongest companies would then need to own the software ecosystem running on top rather than just the onboard computer itself.

Geospatial Intelligence & Applications

What is this category about?

• This category is the application layer of the geospatial stack. The companies here do not primarily sell access to raw observations. They turn location based data into an answer that can be used inside a business workflow. The output might be a risk score or an alert. What matters is that the customer no longer needs to interpret the underlying geospatial data themselves.

• The main pain point is that raw geospatial data rarely answers the question a customer actually has. An insurer does not want to inspect aerial imagery simply to see a roof. It wants to know whether that roof changes the probability of a claim. An infrastructure operator does not need another satellite image of a power line. It needs to know whether vegetation is becoming a threat. These products translate observations into the language of the decision.

• Another problem is that many important physical risks change continuously while the workflows used to manage them are still periodic. Flood exposure can change during an event. Construction projects progress between site visits. Commodity production develops across a season. Geospatial intelligence allows companies to monitor these changes remotely instead of relying on occasional inspections or manually updated databases.

• The category also solves an evidence problem. More business decisions now require companies to prove what happened at a specific location. This is particularly important when the decision affects underwriting or regulatory compliance. Geospatial intelligence can create a repeatable evidence layer that is easier to audit than relying on manual reporting from the field.

What do products in this category do?

• Insurance is one of the clearest verticals for geospatial intelligence because physical risk is inherently location specific. ZestyAI uses aerial imagery and property data to help insurers assess risks such as wildfire exposure or roof condition before underwriting a property. Floodbase focuses on flood risk and goes further by embedding geospatial intelligence into insurance products themselves. Here, geospatial data turns physical conditions that were previously difficult to observe into variables that can directly affect pricing and coverage.

• Infrastructure and energy companies use geospatial data to understand what is happening across assets that are too large to inspect manually. LiveEO monitors infrastructure networks from satellite imagery so operators can detect vegetation risks or other changes remotely. Plume applies geospatial intelligence earlier in the infrastructure lifecycle by helping renewable energy developers evaluate potential sites. In this vertical, geospatial data reduces the need for repeated field inspection and makes large physical networks easier to manage.

• Agriculture uses geospatial intelligence to turn differences across land into operational decisions. OneSoil analyses fields from satellite data and converts the results into tools such as variable rate application maps that farmers can use with their machinery. Treefera works higher in the agricultural supply chain by monitoring the origin and condition of commodities. The value comes from understanding land at a much finer level than farm wide averages or supplier reported information allow.

• Mining and natural resources use geospatial data to decide where physical exploration should happen. TerraEye analyses satellite observations to identify areas that are more likely to contain relevant mineral deposits before expensive field work begins. Kayrros uses remote observations in energy markets to estimate activity and emissions around physical assets. Geospatial intelligence therefore becomes a way to narrow uncertainty before companies commit capital on the ground.

• Maritime industries use geospatial intelligence because much of their operating environment is difficult to observe continuously. SynMax combines satellite data with other signals to track energy related maritime activity and infrastructure. Ocean Intelligence applies environmental observations to aquaculture and other coastal operations. In both cases, geospatial data provides visibility over areas where direct monitoring would otherwise be expensive or intermittent.

What types of customers are using these products?

• Insurance is one of the clearest customer groups because geospatial intelligence can directly affect underwriting. Amica expanded its use of ZestyAI into a broader property risk deployment, while Lemonade adopted the platform for catastrophe underwriting. Floodbase sits closer to the insurance product itself. Its work with Liberty Mutual shows how geospatial flood intelligence can be embedded directly into quoting rather than remaining a separate analytical tool.

• Financial institutions use these products when physical conditions affect the value or risk of an asset. Sust Global worked with Yield Book to bring physical climate risk into mortgage backed securities analysis. Its work with Equitix applied geospatial climate models to wind assets. In these cases the geospatial product becomes another input into an investment decision rather than a mapping tool used by a specialist team.

• Companies exposed to environmental reporting use geospatial intelligence to replace information that is difficult to obtain from suppliers themselves. Maple Credit used Treefera to support forest carbon verification. Kayrros has packaged methane intelligence specifically for financial institutions that need better emissions information about companies they finance. The buyer is paying for independent evidence rather than simply another dataset.

• Infrastructure and energy buyers use these products when physical progress or risk is expensive to verify manually. Kyuden International invested in LiveEO as part of its expansion into Japan and planned a vegetation monitoring demonstration. SynMax is building a similar commercial motion around satellite observed construction progress, where infrastructure buyers can follow projects without relying entirely on reported completion dates.

What are the major trends shaping this category?

• Regulation is making geolocation part of the compliance record. The revised EUDR will apply to large and medium operators from 30 December 2026, and the rules require geolocation information for the land where covered commodities were produced. This creates a structural need for software that can connect supply chain records with independent observations of what happened on the ground. The important shift is that geospatial analysis becomes part of proving compliance rather than an optional sustainability tool.

• Physical risk is becoming harder for insurers to price using historical data alone. Global insured natural catastrophe losses reached $107 billion in 2025, with wildfires and other secondary perils accounting for a record share. Swiss Re estimates that exposure growth explains more than 80% of the long term increase in weather related insured losses. This pushes insurers toward more granular information about individual assets and current conditions, which is exactly where geospatial intelligence becomes useful.

• Defense is becoming a much larger commercial market for geospatial intelligence. NATO's Commercial Space Strategy explicitly calls for greater use of commercial space services in operational planning. In the US, NGA's CIBORG initiative is creating a standardized way to buy commercial geospatial analytical capabilities rather than only imagery. This opens government procurement to companies that sit higher in the stack and sell intelligence products built on top of commercial observation networks.

• AI is changing the unit of geospatial work from analysis toward automated detection. NGA expects the volume of available GEOINT data to triple over the coming decade and is explicitly using AI to generate detections and tip offs from that growing data stream. Humans cannot manually inspect every new observation as collection frequency increases. This makes the application layer more important because software increasingly decides which physical changes are worth putting in front of a user.

How does the funding environment look?

• The strongest companies in this category can attract conventional growth software rounds. Treefera moved from a $12 million Series A in 2024 to a $30 million Series B the following year. ZestyAI raised a $33 million Series B earlier in its development. These are meaningful rounds without the enormous capital requirements of building a satellite constellation, which makes the application layer structurally easier to finance than the sensing layer.

• Funding appears strongest when the geospatial technology is attached to a valuable vertical workflow. Investors are not simply financing companies because they can analyse satellite data. Treefera is building around first mile intelligence for commodity markets, while ZestyAI is embedded in insurance decisions. The commercial story becomes much stronger once geospatial technology disappears behind a problem that already has a large budget.

• Strategic acquisition is becoming a credible outcome for the category. Sust Global was acquired by ISS STOXX in 2025, putting its climate intelligence inside a much larger financial data platform. Kayrros was acquired by Energy Aspects in 2026, giving its geospatial intelligence broader distribution in energy markets. TerraEye was also acquired in 2025. These deals suggest that geospatial applications can become valuable capabilities inside larger vertical information businesses.

• Overall, this looks VC compatible and active, but not currently a hyped market. There is enough capital flowing to support large growth rounds, while recent acquisitions show that strategic buyers see value in the technology. The market remains selective because customers usually buy a specific business outcome rather than generic geospatial AI. That favors startups that become deeply embedded in one workflow over companies selling another horizontal analytics interface.