Google expands Earth AI: Geospatial Reasoning, Google Earth tools, and Cloud access

Google is expanding Earth AI with Geospatial Reasoning, a Gemini-powered framework that fuses multiple geospatial models to answer complex questions, and is opening a Trusted Tester program. New Earth AI imagery capabilities are coming to Google Earth, letting users find objects and patterns (like algae blooms) via natural language, with U.S. availability to Professional tiers and higher limits for AI Pro and Ultra subscribers. On Google Cloud, Earth AI models are available to Trusted Testers, and pilots with WHO AFRO, Planet, Airbus, and Alphabet’s Bellwether show practical impact from disease risk mapping to disaster and infrastructure management.

Key Points

• Geospatial Reasoning links multiple Earth AI models with Gemini to answer complex, real-world questions (e.g., vulnerability and infrastructure risk in storms), with a Trusted Tester program opening now.
• Gemini capabilities in Google Earth gain new Earth AI models so users can query imagery to detect patterns like dried rivers or algae blooms; U.S. rollout to Professional tiers in weeks, with higher limits for AI Pro and Ultra subscribers starting today.
• Earth AI models (Imagery, Population, Environment) are coming to Google Cloud for Trusted Testers, enabling businesses to combine proprietary data with Google datasets such as Imagery Insights.
• Demonstrated impact includes WHO AFRO cholera risk mapping in the DRC, Planet’s deforestation analyses, Airbus’s vegetation encroachment detection, and Bellwether’s hurricane insights for faster insurance claims.
• These advances build on Google’s geospatial AI used in floods, wildfires, cyclones, and air quality, aimed at turning rapid insights into action for environmental monitoring and crisis response.

Sentiment

The community is moderately skeptical. There is genuine interest in geospatial AI as a technology category, and positive engagement around specific use cases like geolocation and LLM-assisted mapping tools. However, significant doubt surrounds Google's corporate framing — particularly the insurance partnership, the timing relative to government agency cuts, and whether the announced capabilities deliver on their promise. The overall tone is cautiously interested but distrustful of the messaging.

In Agreement

• Gemini-in-Google Earth could meaningfully lower the barrier for non-GIS specialists to ask complex geospatial questions
• The technology represents a genuine evolution from static map layers to dynamic reasoning about what, where, and why
• Satellite image search powered by AI would be transformative for insurance, real estate, and disaster response when it matures
• Google's launch of its own climate-monitoring satellite and DeepMind's weather prediction work show real technical progress in this space
• Using LLMs for geospatial analysis is already proving valuable, as demonstrated by the community's enthusiasm for LLM-assisted QGIS and Overpass workflows

Opposed

• The insurance use case is suspect — faster claim processing likely means faster denials, not faster help for homeowners
• Google is effectively profiting from the gutting of government environmental and climate agencies without acknowledging that context
• Existing geospatial datasets like SSURGO soil maps and vegetation indices already serve many of the advertised use cases without requiring AI
• Open-source tools like LLM-generated Overpass queries and QGIS provide similar capabilities without corporate platform dependency
• Current capabilities are overstated — testing shows the system cannot actually perform visual search on satellite imagery yet
• The desire is for taxpayers to fund satellite infrastructure while private companies exclusively profit from selling forecasting