Wind power is intermittent, which makes it less valuable to the grid than energy that can be promised at a set time. On February 26, 2019, DeepMind and Google announced that they had applied machine learning to roughly 700 megawatts of wind power capacity in the central United States to address this problem.
A neural network was trained on widely available weather forecasts and historical turbine data to predict wind power output 36 hours ahead of actual generation. Using those predictions, the system recommends how to make optimal hourly delivery commitments to the grid a full day in advance, when such commitments are worth more than simply selling power as it happens. DeepMind reported that this approach boosted the value of the wind energy by roughly 20 percent compared with a baseline of no time-based commitments.
The case matters because it shows AI improving the economics of renewable energy not by generating more power, but by making existing output more predictable and schedulable - a lever that grid operators and energy traders care about directly.