Font Size: 

The transition from gasoline-powered vehicles to electric vehicles (EVs) presents a promising avenue for reducing greenhouse gas emissions. Spatial forecasts of EV adoption are essential to facilitate this shift as they enable preparation for power grid adaptation. However, forecasting is hindered by the limited data availability at this early stage of adoption. Multiple model calibrations can match current adoption trends but yield divergent forecasts. By leveraging empirical data from places with leading adopters in the US, this study shows that taking into account the spatial and social structure linking potential EV adopters leads to forecasts of only 25\% of the current predictions for 2050.