Drawing on 70 years of historic wind and solar-power data, researchers built an AI model to predict the probability of a network-scale ‘drought,’ when daily production of renewables fell below a target threshold. Under a threshold set at the 30th percentile, when roughly a third of all days are low-production days, the researchers found that Texas could face a daily energy drought for up to four months straight. Batteries would be unable to compensate for a drought of this length, and if the system relied on solar energy alone, the drought could be expected to last twice as long — for eight months.