Climate tech company Tomorrow.io has revealed the design behind DeepSky, a planned satellite constellation built specifically to collect high-frequency weather data for AI-based forecasting models. Instead of treating satellites and AI systems separately, DeepSky is being designed from the ground up as a data source optimized for machine-learning weather prediction.
AI weather models have improved a lot recently, but their accuracy still depends heavily on how much fresh, high-quality data they can access. Better satellite coverage could mean earlier warnings for hurricanes, floods, and other severe weather — which matters most in regions that don’t have strong radar networks already in place.
It’s also a good example of a bigger trend: AI is moving out of software-only products and into physical infrastructure like satellites, sensors, and climate monitoring systems.