A new AI weather forecasting startup outperformed world-leading government-run systems, the latest evidence of the technology disrupting a centuries-old science.
WindBorne Systems’ five-day forecasts are reportedly as accurate as cutting-edge physics models’ one-day predictions, and require far less powerful machines.
Empirical weather forecasting dates back to the 1860s, but AI is increasingly reshaping it: Since a Google DeepMind model beat the European Centre for Medium-Range Weather Forecasts in 2024, the ECMWF has begun incorporating AI into its own forecasts.
The rise of private forecasts is timely, as the US cuts National Weather Service funding, and as forecasts become ever more important to energy projections and trading thanks to the rise of renewables.