Storms Are Coming. Can AI Protect the Power Grid?

Researchers unveil an AI model that blends weather forecasts and outage data to predict grid failures before storms hit

13 Mar 2026

In February a pair of American universities announced an unusual ambition: to predict power cuts before the wind starts to blow. Researchers at the University at Albany and the University of Connecticut say artificial intelligence may help electricity utilities see where storms will break the grid, hours or even days in advance.

Their project, the North American Forecasting Model, is being developed through the universities’ joint WISER research centre. The idea is simple, though the data challenge is not. The system blends high-resolution weather forecasts with historical records of power outages gathered from utilities across the continent. By learning how particular storm conditions translate into broken lines and fallen equipment, the model aims to estimate where failures are most likely to occur.

Utilities have long tried to anticipate outages. Yet most tools rely on local data collected by individual companies. The new effort seeks something broader: a system trained on outage histories spanning North America. If it works, operators could position repair crews before a storm hits rather than dispatching them after the damage is done.

The project has attracted modest but notable backing. An industry advisory board of large American utilities, including operators in the north-east and on the west coast, has committed $550,000. The National Science Foundation, which established the WISER centre in 2023, provides the research programme’s underlying support.

Testing will begin close to home. The first phase will pilot the model across New England, New York and California. Expansion to other parts of the United States and Canada will depend on further funding and data-sharing agreements.

The interest is hardly academic. American electricity customers are paying steadily more for reliability. Regulators approved $11.6bn in rate increases across 43 utility cases in 2025 alone. In New York state, electricity bills have climbed by roughly a third over the past decade. Faster restoration after storms could reduce the operational costs that utilities ultimately pass on to consumers.

Extreme weather is becoming more common even as much of the grid grows older. That combination has made “predictive resilience” a fashionable phrase in energy circles. Whether algorithms can truly foresee the next fallen power line remains uncertain. But utilities are increasingly willing to try.