The rapid electrification of the transportation fleet imposes unprecedented demands on the electric grid. If controlled, however, these electric vehicles (EVs) provide an immense opportunity for smart grid services that enable renewable penetration and increased reliability. In this talk we discuss paradigms for aggregating and optimally controlling EV charging. Specifically, we discuss (i) aggregate modeling via partial differential equations, (ii) distributed robust optimization of large-scale EV fleets, (iii) and plug-and-play model predictive control. The talk closes with future perspectives for EVs in the Smart Grid.