New AI Maps Best Places for Solar Farms Worldwide
This could make solar power cheaper and faster to build, saving you money on energy bills.
Researchers used geospatial embeddings—dense vectors representing geophysical properties of a location—to analyze 38,591 existing solar sites and the land they were built on, learning the latent solar suitability signal. Their soft-voting ensemble of a random forest and a multilayer perceptron was validated against future installations in thirteen countries, separating existing solar sites from random land with a median ROC-AUC of 0.90 and increasing ROC-AUC by 0.06–0.30 compared with published MCDA strategies in every country.
The ranking concentrated real deployment: the most-suitable 20% of land held 84% of estimated generation and 67% of installed capacity. After applying standard siting constraints, 5.6% of global land remained highly suited to solar—9.8% if cropland was permitted.
- AI analyzes satellite data to find the best solar farm locations, making solar energy cheaper and faster to deploy.
- Only 5.6% of global land is highly suitable for solar, but that's enough to power the world many times over.
- The tool outperforms traditional methods in 13 countries and is free for anyone to use.
Why It Matters
Faster solar farm siting means more clean energy, lower bills, and new jobs in renewable energy.