From Floods to Droughts: The New Water Resilience Stack

Climate change is not only changing how much water is available. It is making the whole water cycle less predictable. Rain can arrive too late, too heavily, or in the wrong place, while long dry periods make existing water reserves less reliable. I wanted to look at the startups building around this new reality, from better forecasting and flood intelligence to groundwater monitoring and systems that capture excess rain so it can be used later.

Water Observation & Forecasting

What is this category about?

• Water observation and forecasting products try to give users a more accurate picture of what is happening locally than they can get from a general weather forecast. Rainfall can vary significantly across short distances, while many operational decisions depend on what happens at one specific location.

• Observation is still a major limitation. Weather stations provide direct measurements but are unevenly distributed. Radar and satellites provide broader coverage, but the information still needs to be converted into a reliable local estimate. The companies in this category use software or new observation networks to close that gap.

• The second problem is timing. Knowing that rain is expected tomorrow is not enough for many users. They need to know when it will start, how much will fall and whether conditions are changing faster than the normal forecast cycle can capture.

• Some products go beyond forecasting rain and predict what that rain will do to rivers and local water levels. This matters because 50 mm of rain does not create the same flood risk everywhere. The impact depends on the river, the terrain and how much water is already in the system. These products model that next step so users can forecast river discharge or flooding rather than rainfall alone.

What do products in this category do?

• One product model improves local precipitation forecasts using existing weather data. The logic is that customers do not need another global weather model. They need a better answer for a particular location. Precip combines weather observations with forecast models to estimate rainfall and snowfall at a local level. Rainbow Weather focuses more heavily on minute-by-minute nowcasting and exposes the forecast through APIs that other products can integrate.

• A second approach improves the observation layer itself. Forecast quality depends on knowing what is happening now, but traditional station networks leave large gaps. WeatherXM tackles this by building a distributed network of privately operated weather stations. The resulting observations are then packaged as a data product rather than remaining tied to the hardware.

• WeatherXM is not really innovating on the basic weather station itself as much as on how thousands of relatively cheap stations become a reliable network. A station measures rainfall, temperature, wind and other local conditions, then sends readings through Wi-Fi, LoRaWAN or 4G. Current hardware ranges from roughly $139 for the discounted Wi-Fi/Helium models to $810 for the 4G version. The more interesting layer is the network: WeatherXM verifies each station’s location, scores the quality of its data against neighbouring stations and models, and rewards owners for contributing useful observations. This makes it possible to build dense ground coverage without WeatherXM deploying every station itself.

• The third model turns weather information into forecasts about water conditions. This exists because governments or infrastructure operators often care more about river behaviour than rainfall itself. Aqunia combines meteorological inputs with hydrological models to forecast river discharge and flood conditions. The product therefore sits one step further down the decision chain than a normal weather API.

• AI is improving this category, but it is mostly accelerating an existing move toward more local and frequently updated forecasts. Weather models were already becoming more precise before the current AI wave. AI makes it cheaper to combine large amounts of weather data and update predictions quickly, but the harder problem is still getting good local observations. This is why companies such as WeatherXM can create value simply by improving the underlying data available to the models.

What types of customers are using these products?

• Agriculture is one of the clearest commercial use cases because small differences in rainfall can change operational decisions. OneSoil integrated Rainbow Weather forecasts into its farming platform, where the feature was already being used by more than 15,000 farmers in June 2026. Verdi embedded Precip weather data inside its irrigation software so customers could use local rainfall information directly inside an existing workflow.

• Public agencies use these products when they need forecasting in places where local hydrological information is limited. Aqunia is working with authorities in Gia Lai, Vietnam on flood early warning. It was also selected for a JICA-backed project in Zambia that uses satellite rainfall estimates and hydrological modelling for national water-resource planning.

• Another route to market is to become the weather data layer inside another product. WeatherXM has worked with TRUF.NETWORK on weather-linked financial products. Its SwissBorg rollout also helped deploy 2,270 stations across 19 countries. These partnerships matter because the weather company does not need to acquire every final user directly.

What are the major trends shaping this category?

• AI weather models are becoming good enough to be used operationally. ECMWF (the European Centre for MediumbRange Weather Forecasts, Europe’s main weather forecasting organisation) started using an AI forecasting system alongside its traditional physics based model in 2025. This makes strong global weather forecasts cheaper and easier to produce. The opportunity for startups may therefore move toward improving forecasts for a specific location rather than building another global model.

• Good forecasts still depend on good observations. WMO (the World Meteorological Organization, the UN body for weather and climate) says many parts of the world still lack enough basic weather observations. This is especially true in poorer countries. Better models cannot fully compensate for missing information about what is actually happening on the ground, which creates an opportunity for new sensor networks.

• Extreme rainfall is becoming more intense as the climate warms. The IPCC (the UN scientific body that assesses climate change) estimates that extreme daily rainfall increases by roughly 7% for every degree of global warming. This makes accurate local forecasts more valuable because a small forecasting error matters much more during an extreme event.

• The interesting part of the market may increasingly sit between global weather models and the final user. Global forecasts are becoming easier to access, but businesses still need to know what will happen at their exact location and what it means for their operations. Startups can create value by improving local data or turning general forecasts into something directly useful for a specific decision.

How does the funding environment look?

• Overall, the category looks moderately VC compatible rather than hot. Weather data can become a high margin API product and proprietary observation networks can create a real data advantage. But the category has not produced a broad funding wave, and several companies remain very small. The current dynamic looks early and active rather than hyped.

• The largest venture funding has gone to companies that can turn weather data into a scalable infrastructure product. WeatherXM raised a $5 million seed round followed by a $7.7 million Series A. Its investment case is partly about forecasting, but also about building a proprietary observation network that becomes more useful as more stations are deployed.

• Rainbow Weather shows that an API first forecasting company can also attract meaningful capital once distribution is visible. It raised $5.5 million in 2026 after an earlier €1.8 million round. At the time of the latest financing, the company had passed one million installs and later reported 400 B2B API clients.

• Public financing is more important when the product moves closer to hydrology and government deployment. Aqunia has received support through Japanese government programmes rather than announcing a conventional venture round. Its projects require local validation and public-sector partnerships, which makes grants a natural source of early funding.

Flood Intelligence & Response

What is this category about?

• Flood intelligence products try to show where flooding will happen and what it will affect before the water arrives. Traditional flood maps are useful for long term planning, but they are often too static for an event that changes hour by hour. The products in this category add live conditions or forecasts to make flood risk operational.

• The difficulty is that rainfall does not translate directly into flooding. The same amount of rain can have very different consequences depending on the terrain or how saturated the ground already is. In cities, drainage capacity can completely change the outcome. This makes local modelling important.

• Timing is another major problem. Knowing that a property is inside a flood zone does not tell an emergency manager whether it will flood tonight. Products therefore increasingly combine long term risk information with short-term forecasting so users can move from understanding exposure to anticipating an event.

• The final value is not the flood map itself. Emergency teams need to decide where to intervene. Insurers need to understand which assets are exposed and whether an event has crossed a predefined threshold. The product becomes more valuable when the flood model is connected directly to one of these decisions.

What do products in this category do?

• One model forecasts flooding before and during an event. The logic is to translate incoming rainfall into the streets or properties that are likely to be affected. FloodMapp provides forecasts before an event and live inundation maps as conditions develop. Previsico follows a similar model with probabilistic forecasts that update hourly, while its platform adds monitoring information that helps customers confirm what is happening on the ground.

• A second group turns flood modelling into a persistent risk layer for financial decisions. These products exist because an insurer or bank needs to evaluate thousands of properties even when no flood is happening. 7Analytics produces property-level flood risk data that can feed underwriting or credit workflows. Floodbase has taken this further into insurance infrastructure, where its flood data can define parametric triggers and now power automated quoting through an API.

• Another approach builds the observation infrastructure needed for local early warning. Models become less useful when there is little ground data available. Green Stream deploys waterblevel sensors and connects them to a cloud platform for alerts. Aurassure uses a broader environmental sensor network but applies the same model to urban flooding, including a citywide deployment connected to Chennai's command centre.

• StormHarvester applies flood intelligence inside the wastewater network itself. The product exists because urban flooding can begin with overloaded sewers or a pump problem rather than a river leaving its banks. StormHarvester uses live network data and rainfall information to detect problems before they become incidents. Yorkshire Water now uses the platform across more than 4,400 sensors, while South West Water expanded it from a trial to roughly 1,400 overflow points.

• AI is having a more visible impact here because it helps turn flood modelling into a live operational product. Flood models have existed for decades, but they were often slow to run or used mainly for planning. New systems can combine forecasts with sensor data and continuously update which locations are likely to flood. AI is therefore not creating flood modelling from scratch, but it is helping move the category from static risk maps toward systems that can support decisions while an event is happening.

What types of customers are using these products?

• Governments and emergency-management agencies are a major customer group. FloodMapp was used by the Texas Division of Emergency Management during the 2025 floods across more than 22 counties. The City of Boston also bought its forecasting products for emergency planning. Green Stream works with regional authorities in Hampton Roads, where its network covers 15 communities.

• Insurers and financial institutions use flood intelligence before an event rather than only during one. Floodbase works with companies such as Liberty Mutual to embed parametric flood quoting directly into insurance workflows. 7Analytics supplies property-level climate risk to Nidaros Sparebank for credit decisions. Previsico has also distributed its forecasts through insurance relationships including Zurich.

• Water utilities and infrastructure operators use these products to protect physical networks. StormHarvester has become deeply embedded in UK wastewater operations. Northumbrian Water awarded it a long-term contract covering a 30,180 km network, while Yorkshire Water uses it to prioritize interventions. National Grid uses Previsico flood forecasts for infrastructure sites exposed to surface water flooding.

What are the major trends shaping this category?

• Flooding is becoming harder to manage because extreme rainfall is getting more intense. The IPCC (the UN scientific body that assesses climate change) expects heavy rainfall to become stronger in many regions as the planet warms. At the same time, more people and infrastructure are concentrated in cities, where drainage systems can be overwhelmed quickly. This increases the value of tools that can show where flooding is likely to happen before the water arrives.

• Flood warnings are becoming more focused on consequences rather than just the weather itself. WMO (the World Meteorological Organization, the UN body for weather and climate) is encouraging what it calls “impact based forecasting.” Instead of only warning that a river will rise or that heavy rain is coming, the aim is to predict which roads, buildings or communities are likely to be affected. This is very close to what many startups in this category are building.

• Flood data is increasingly being used directly inside insurance products. A large share of flood losses remains uninsured, which is pushing insurers to experiment with products that can pay automatically when a flood reaches a predefined level. This creates demand for flood measurements and models that are reliable enough to determine whether a payout should happen.

• Basic flood maps are becoming easier to access for free. Copernicus (the European Union’s Earth observation programme) already provides satellite based flood maps during major disasters. This means startups need to offer more than a map. The stronger products provide earlier forecasts, more local detail or connect the information directly to decisions such as evacuations, insurance payouts or infrastructure operations.

How does the funding environment look?

• I would describe the category as VC compatible and active, but not hyped. There are already companies with meaningful venture rounds and real evidence of scale. StormHarvester reported more than 3,000% four year revenue growth, while Floodbase has supported more than 9,000 insurance policies. The constraint is that many customers are utilities or public agencies, where adoption takes longer than in normal software markets. That makes the category attractive, but unlikely to produce the kind of funding wave seen in AI infrastructure.

• The largest rounds have gone to companies that connect flood intelligence to a valuable operational workflow. Floodbase raised a $12 million Series A and later added $5 million in follow-on financing as it expanded deeper into insurance infrastructure. StormHarvester raised £8.4 million in 2025 to scale its wastewater intelligence platform. Both sell more than a risk score: their software becomes part of how customers operate.

• Strategic investors are appearing as the category matures. IAG Firemark Ventures invested in 7Analytics as the company expanded its insurance business. Previsico has attracted investors including Burnt Island Ventures and BlueOrchard, while its most recent financing supports expansion into Africa. These investors have a direct interest in climate resilience or insurance rather than treating flood software as generic SaaS.

• Public funding still plays an important role in getting products deployed. FloodMapp used resilience funding in Virginia before expanding into a broader regional rollout. Previsico has received grant backing for geographic expansion, while Green Stream began with support from NC IDEA. This matters because public sector deployments can take time to reach normal commercial scale.

Drought, Groundwater & Water Availability Intelligence

What is this category about?

• Groundwater is difficult to manage because most of the resource is invisible. A river can be observed directly, but understanding what is happening underground requires wells, sensors or models. This makes it difficult to know how much water is available and whether an aquifer is being depleted.

• The available information is often incomplete or updated too slowly. Groundwater may be measured at individual wells only a few times per year, while conditions can vary significantly between locations. This becomes a bigger problem during long dry periods, when users need to understand whether water reserves are actually recovering.

• The category exists to make water availability easier to measure before it becomes a crisis. Some products monitor groundwater continuously. Others help users find water before drilling or estimate conditions across much larger areas. The common goal is to replace assumptions about water availability with more current information that can support planning.

What do products in this category do?

• One product model improves what can be measured directly underground. Traditional groundwater monitoring often relies on individual wells and periodic measurements. iFLUX installs sensors that continuously measure how groundwater moves, including its speed and direction. G-Strata addresses an earlier part of the problem: it uses geophysical measurements to understand underground formations before a new well is drilled. This can reduce the risk of drilling in the wrong place.

• Another approach uses satellites and models to understand water conditions across much larger areas. This is useful when installing sensors everywhere would be too expensive. HydroGeoTwin combines satellite observations with local measurements and hydrological models to estimate groundwater recharge and future availability. Hydro-EO follows a similar logic for broader water monitoring. PlanetAI Space uses satellite information together with existing well data to identify changes in groundwater conditions.

• A different model turns water availability into business risk. Large companies may operate hundreds of sites without having a consistent way to know which ones face growing water stress. Waterplan combines local water data with company information so businesses can identify exposed facilities and decide where action is needed. The product is therefore less about discovering groundwater and more about making water scarcity manageable across a large organisation.

• AI is useful in this category, but it does not remove the need to measure water. The main problem is still that groundwater data is sparse and comes from many different sources. AI can combine sensor readings with satellite information or help identify patterns in historical data. It can make the final estimate better, but a model cannot compensate indefinitely for having no information about what is happening underground.

What types of customers are using these products?

• Water utilities are natural customers because groundwater is part of their core supply. Vitens uses an iFLUX sensor network around its Epe extraction site in the Netherlands to understand how groundwater moves. Belgian utility Pidpa extended its work with iFLUX for another four years after an initial monitoring project.

• Cities and public water managers use these technologies when they need to find or manage new water sources. The municipality of Vilcún in Chile purchased groundwater prospecting work from G-Strata for drinking-water supply. iFLUX is also being used in Ghent to understand groundwater movement beneath Citadelpark as the city develops a system designed to store more rainwater underground.

• Large companies use water intelligence to understand which sites could face supply problems. Waterplan names Amazon, AB InBev, Colgate Palmolive and Coca Cola Europacific Partners among the large organisations using its platform. PlanetAI Space says it works with customers in sectors such as mining and energy, where access to water can directly affect whether an asset can continue operating.

What are the major trends shaping this category?

• Water availability is becoming less predictable. WMO (the World Meteorological Organization, the UN body for weather and climate) found that only around one third of the world's river basins had normal conditions in 2024. The rest had either more or less water than usual, continuing a multi-year pattern of unusually unstable water conditions. This makes historical averages less useful for planning and increases the value of more frequent monitoring.

• Groundwater levels are already falling quickly in many places. A large study published in Nature analysed measurements from 170,000 wells and found widespread groundwater decline, particularly in dry agricultural regions. In 30% of the aquifers studied, the decline had accelerated over the previous four decades. This makes groundwater monitoring less about finding a static reserve and more about understanding how quickly that reserve is changing.

• Satellites are making it possible to monitor water in places with few ground measurements, but they cannot replace local data. NASA (the US space agency) can estimate regional changes in underground water using satellites that detect changes in Earth's gravity. The limitation is resolution: the measurements cover very large areas and cannot tell a utility exactly what is happening around one well. This explains why many products combine satellite information with sensors or hydrological models rather than relying on space data alone.

• Water scarcity is becoming a business risk rather than only an environmental issue. CDP (a global environmental disclosure platform used by companies and investors) says companies reported $339 billion of potential financial impacts linked to water risks in 2025. More businesses are therefore trying to understand water conditions around individual factories and suppliers before shortages interrupt operations. This creates a larger market for products that translate hydrological information into site-level business risk.

How does the funding environment look?

• Overall, the category looks moderately VC compatible but still slow rather than trending. Enterprise software such as Waterplan can fit a traditional venture model well. Groundwater sensing and exploration are harder because deployments require physical work and specialist buyers. The underlying problem is becoming more important, but the current funding pattern does not suggest a broad investor rush into the category.

• Waterplan is the clear venturebacked outlier in the category. It raised a $7 million seed round followed by an $11 million Series A. Its model is also the closest to conventional enterprise software: one platform can be sold across many sites inside the same large company without requiring new hardware to be installed at each location.

Stormwater Capture, Recharge & Buffering

What is this category about?

• Most cities were designed to get rainwater away as quickly as possible. Water falls on roofs or roads and is sent into drains. During heavy rain, this can overwhelm the drainage system. The same city may then face water shortages a few months later.

• The companies in this category try to keep more of that water locally. Rain can be temporarily stored instead of immediately entering the sewer system. It can then be reused later or allowed to recharge groundwater.

• The difficulty is that storage needs to work across very different weather conditions. A tank that is already full cannot absorb the next storm. Water that will be reused may also need treatment. This creates room for systems that do more than simply collect rain in a large container.

What do products in this category do?

• One approach builds complete systems for capturing rain and storing it for later. The logic is to solve two problems with the same infrastructure: reduce pressure during heavy rain while keeping that water available during dry periods. FieldFactors' BlueBloqs captures stormwater, treats it and stores it underground. At its GreenLED site in the Netherlands, around 600 cubic metres can be buffered and then stored seasonally in an aquifer for later irrigation.

• A second subcategory of startups distributes smaller storage systems across individual properties. This matters because a large part of urban runoff starts on private roofs rather than public land. RainGrid's system redirects roof runoff into local storage, where the water can be reused instead of immediately entering the sewer. The broader idea is that many small installations together can create meaningful storage across a neighbourhood.

• Another approach makes existing stormwater infrastructure more intelligent rather than building more storage. OptiRTC connects weather forecasts with water level sensors and automated valves. If a large storm is approaching, the system can release some stored water beforehand so that capacity is available when the rain arrives. The same pond or tank can therefore handle more water without physically becoming larger.

• AI is not fundamentally changing this category yet. The main innovations are still better physical systems and the ability to control them automatically. Better forecasting can help decide when water should be stored or released, and AI may improve those decisions over time. But unlike flood forecasting, the main bottleneck is not the model. It is building enough storage and connecting it to the places where rain actually falls.

What types of customers are using these products?

• Property developers and large companies can use these systems to manage rainwater directly on their sites. FieldFactors is working with Microsoft on a project in Madrid that aims to capture stormwater and replenish the nearby Valdebebas stream. It has also piloted BlueBloqs at an AB InBev facility in Mexico City.

• Cities are another natural customer because stormwater is normally a municipal problem. FieldFactors' operational project in Alphen aan den Rijn shows how captured rain can reduce pressure during wet periods and then provide irrigation water during dry ones. This gives cities an alternative to treating flood prevention and water supply as completely separate systems.

• Existing infrastructure companies can also become distribution partners. Oldcastle Infrastructure launched SmartCapture in 2026 with Opti's control software built into its stormwater storage system. This is an interesting route to market because the software company can reach new projects through an established infrastructure supplier rather than selling every installation itself.

What are the major trends shaping this category?

• Cities are looking for alternatives to simply building larger drains and pipes. The US Environmental Protection Agency (the federal environmental regulator) says much existing stormwater infrastructure is ageing and is often too small for the amount of water it now needs to handle. Capturing water where it falls can reduce the amount entering the drainage network during a storm.

• Stormwater is increasingly being treated as a water source rather than waste that needs to disappear. The US Environmental Protection Agency now explicitly promotes captured stormwater for reuse. Depending on how it is treated, the water can be used locally or stored underground for later. This changes the economics of stormwater infrastructure because the same investment can reduce flooding and create a new water supply.

• Storing water underground is becoming more interesting as a way to connect wet periods with dry ones. UNESCO (the UN agency responsible for areas including science and water research) has documented projects where excess water is deliberately put back into aquifers so it can be available during droughts. Underground storage also avoids some of the evaporation that occurs in surface reservoirs.

• Software can increase the capacity of infrastructure that already exists. The US Environmental Protection Agency highlights the example of Ormond Beach in Florida, where forecast driven controls created space in a lake system before a hurricane. The control system cost about $200,000, compared with an estimated $8 million for additional pumping infrastructure. This makes smart control particularly interesting for cities that cannot simply rebuild their entire drainage network.

How does the funding environment look?

• Overall, this looks less VC compatible than the other categories in the landscape. 

• FieldFactors is the clearest venture backed company in the category, but the rounds remain small. It raised €1.65 million in equity in 2025 and another undisclosed round in 2026. The company is using the capital to move from demonstration projects toward broader international deployment.

• Public funding is unusually important because projects require physical infrastructure before the business can scale. FieldFactors participates in a €3 million European Union LIFE project that helped finance real world demonstration sites. RainGrid received a CAD 250,000 co-investment after winning a World Economic Forum water challenge.

• Opti shows another possible outcome for the category: becoming part of a larger infrastructure company. Aliaxis acquired the company in 2022, although it reported that Opti had been divested again during the first half of 2026. The acquisition still shows why control software can be strategically interesting to manufacturers that already sell physical water infrastructure.