Transform planetary information into actionable intelligence

Specialized AI models enable finding objects, classifying scenes, and monitoring changes across Earth's observation imagery. Query the Earth with natural language and significantly accelerate remote sensing workflows, such as identifying where harmful algae is blooming in order to monitor drinking water supply, giving authorities time to issue warnings or shut down water utilities.

AlphaEarth Foundations integrates petabytes of Earth observation data to generate a unified representation, empowering users to perform faster, more accurate geospatial analyses like change detection, map segmentation and classification.

Gemini capabilities in Google Earth now integrate new Earth AI models, allowing users to instantly find objects and discover patterns across satellite imagery. This experimental feature streamlines environmental monitoring—such as spotting dried rivers or algae blooms—for Professional and Professional Advanced users in the US.

Create, analyze, and collaborate with Google's unique imagery and global knowledge, cloud infrastructure, and community-scale generative building and solar designs. Make faster, informed decisions in early-stage projects—no GIS training required.

Vantor is integrating Google Earth AI imagery models into its Tensorglobe platform to help government and commercial organizations respond to a rapidly changing world with greater clarity and confidence. By combining Earth AI imagery foundation models with Vantor’s AI-ready spatial foundation—built on a 20+ year satellite imagery archive and their advanced global 2D and 3D data—organizations can fuse vast volumes of sovereign and third-party imagery into timely, actionable insight.

ForestCast, the first deep learning benchmark for proactive deforestation risk forecasting, is a model that utilizes pure satellite data to predict future forest loss accurately and at scale, overcoming the limitations of older methods that relied on inconsistent, region-specific input maps. This marks a fundamental shift from monitoring past losses to actively predicting and preventing future environmental threats.

Natural Forests of the World 2020, an AI-powered baseline map for deforestation and degradation monitoring. This critical resource achieves best-in-class accuracy at 10-meter resolution in distinguishing natural forests from other tree cover, aiding companies (e.g., for EUDR compliance), governments, and conservation groups worldwide.

This open dataset maps 1.8 billion buildings in the Global South and includes changes in development from the years 2016-2023. It uses AI to super-resolve and extract building footprints and heights from Sentinel-2 satellite imagery. For example, it's allowed us to greatly expand coverage of our Solar API.

Together, Aerial & Satellite Insights and Street View Insights put global, high-resolution imagery at your fingertips. It expands the capability from the fine details of a city street to a bird’s-eye view of the planet, empowering you to make data-driven decisions at any scale. In addition, Aerial & Satellite Models are designed to transform your geospatial workflows by enabling you to extract planetary-scale insights without the need to train or maintain complex models yourself.

Our AI flood forecast provides reliable riverine flood warnings up to seven days. Flood Hub currently covers river basins in over 100 countries worldwide, providing critical flood forecasting to a population of 700M people.

WeatherNext is an AI-powered ensemble forecasting model for global weather prediction. It utilizes a novel Functional Generative Network architecture, which enables it to generate forecasts 8x faster and with resolution up to 1-hour. This state-of-the-art model delivers more accurate global weather predictions - including for extreme weather events - aiding enterprises, governments, and researchers worldwide.

Weather Lab features our latest experimental AI-based tropical cyclone model, based on stochastic neural networks. This model can predict a cyclone’s formation, track, intensity, size and shape — generating 50 possible scenarios, up to 15 days ahead.

Camera traps generate vast amounts of data, but manually identifying millions of wildlife photos can take decades and creates a major bottleneck for conservation research. One year after its open-source release, the SpeciesNet model is empowering researchers to classify nearly 2,500 animal species—processing massive datasets from the savannas of Tanzania to the forests of Idaho—streamlining global biodiversity monitoring and helping conservationists quickly translate raw imagery into actionable insights to protect threatened populations.

For years, a lack of high-fidelity historical data prevented AI models from predicting flash floods before they happen. Groundsource is a new AI-powered methodology that transforms public information into a high-quality record of historical disaster data – starting with flash floods in urban areas. Groundsource uses Gemini to analyze decades of public reports and identifies over 2.6 million historical flood events spanning more than 150 countries. It then used Google Maps to determine precise geographic boundaries for each event to create a dataset focused on flash floods. Using this dataset we trained a new model now available on Google’s Flood Hub to predict flash floods in urban areas up to 24 hours in advance, to help ensure no one is surprised by a natural disaster.

Google’s novel tools aimed at helping us better understand and protect Earth's biosphere include a benchmark dataset for high-resolution deforestation risk prediction, a new approach for species range mapping at unprecedented scale, and Perch 2.0, a state-of-the-art bioacoustics model for automated ecosystem health monitoring and identifying endangered species.

NeuralGCM is a hybrid atmospheric model combining traditional physics with machine learning for faster, accurate weather simulations. By using physical laws for large-scale dynamics and ML for small-scale phenomena, it produces high-quality forecasts at a fraction of the cost. This open-source model significantly increases accessibility for researchers.

Powering air quality information for millions of users on Google Search and Maps, our advanced Air Quality API uses AI to fuse satellite, weather, and traffic data, delivering highly accurate, real-time Air Quality Index (AQI) forecasts at a 500-meter global resolution.

Our Pollen API can be used to help people limit the risks of exposure to allergenic pollen and make better informed daily decisions. The pollen model combines AI with physical and biological modeling to generate a 5-day high-resolution forecast across over 65 countries. The model is trained to recognize the exact location and density of specific tree, grass and weed species to predict the Universal Pollen Index (UPI).

We use satellite imagery and ML to detect and track wildfires, making information available via Google Search and Maps, informing affected communities and helping fire authorities take action, and developing a simulator to generate data in a range of wildfire scenarios.

Measles outbreaks are rising, but routine vaccination data are often delayed and too coarse to pinpoint local gaps. Researchers at Harvard, Mount Sinai, and Boston Children’s Hospital combined survey data with Google’s Population Dynamics Foundation Model (PDFM) to produce “superresolution” estimates of MMR coverage among young children—at much finer geographic scales than standard reporting, down to the ZIP-code level—revealing clusters of undervaccination that align with recent outbreaks and helping public health teams prioritize locally targeted outreach and prevention efforts.

Traffic Simulation API leverages AI advancements in measurement, simulation, and optimization to provide transportation agencies with powerful tools for baseline simulation of city networks, allowing them to modify and test the network-wide impact of proposed infrastructure changes, as well as temporary disruptions such as lane or road closures due to maintenance, mass events, or crashes. It uses high-fidelity modeling to provide granular results on vehicle counts and average speeds, giving users detailed data on the cascading effects across the entire road network to de-risk investments.

Google's Population Dynamics Foundation Model (PDFM) distills data about human behavior and our interactions with the environment into concise, analysis-ready embeddings. Now expanded to include more countries and capture changes over time, PDFM insights are enhancing applications including disease modeling, mapping vulnerable communities and location-based marketing.

Mobility AI leverages AI advancements in measurement, simulation, and optimization to provide transportation agencies with powerful tools for data-driven policy making, traffic management, and continuous monitoring of urban transportation systems. Powering Roads Management Insights on Google Maps Platform.

By integrating real-time disruption data with operational and volume metrics, organizations can gain a comprehensive view of road conditions and safety. Real-time disruption feeds provide immediate alerts on road closures, incidents, and environmental hazards. Complementing these safety signals, vehicle counts offer deep analytics on traffic volume, allowing for more precise maintenance scheduling and sophisticated pattern analysis.

Population Dynamics Insights is a first-of-its-kind geospatial embeddings dataset designed to help organizations decode the complex relationship between human behavior and the physical world. By distilling anonymized trends from Google Search and Google Maps into rich 330-dimensional vectors for locations across the globe, it enables a new era of spatial machine learning without the need for manual feature engineering.