Globally, large mammals are in decline. Biological traits including low population densities and longer generation times make them particularly susceptible. Their losses can have wide-ranging ecological consequences, including dramatic reductions in total heterotrophic energy use. To determine the key drivers of variation in energy use, we calculated daily rates of energy flow across the globe for 241 ecological communities, encompassing 441 large mammal species, using camera trap inventories. These were scaled up from individual metabolic rates and compared with various climate, anthropogenic, geographic, and species richness variables using three analytical methods: model selection, spatial autoregression, and a multiple regression method that completely removes multicollinearity known as least-squares orthogonalization. Community energy use is significantly lower in the Neotropics and Australasia than in the Afrotropics and Eurasia. This pattern mirrors the spatial distribution of megafaunal extinction intensity during the Late Pleistocene. Rates not being greatly reduced in the Nearctic is a notable exception to this pattern, and is likely due to the high abundances of certain species not present in the other highly-impacted realms. There are also strong negative correlations between community per-gram rates of energy flow and species richness, indicating that megafauna persist mainly in more speciose communities. The strong geographic differences that dominate energy use patterns indicate that past mammal extinctions are the ultimate cause of modern energetic variation in large mammal communities. If so, then ongoing losses of large mammals will greatly impact community and ecosystem functioning.