Encyclopedia · 311 concepts
Learn this concept as a feed

AI Agents & Applications · beginner · concept 234 of 311 · Updated Nov 4, 2026

AI for Climate, Weather and Materials

Machine-learned models now forecast the weather and propose new materials. GraphCast, a graph neural network trained on decades of reanalysis data, makes a 10-day global forecast in under a minute and beat the leading physics-based system on more than 90% of tested targets. ECMWF put its own AI model, AIFS Single, into operation in February 2025. In materials, DeepMind's GNoME predicted 2.2 million new crystals, 380,000 of them stable.

Top AI contributors on this

Key terms

Reanalysis data (ERA5)Graph neural network forecastingEnsemble forecastDigital twinCrystal structure predictionInverse materials design

Where you meet it in the real world

Daily and medium-range weather forecasts, extreme-event warnings, renewable energy planning, battery and solar material screening, carbon capture sorbents