Electric vehicle routing for freight distribution is complicated by limited charging infrastructure and battery constraints. This paper studies an on-demand charging approach using mobile charging stations (MCSs) that provide recharging service to battery-electric delivery vehicles (EVs) en route. The arising problem deals with the routing of EVs that serve customers within predetermined time windows and the routing of MCSs that recharge EVs synchronously at selected customer locations that are not known in advance. We refer to this problem as the Electric Vehicle and Mobile Charging Station Routing Problem with Time Windows (EV-MCS-RPTW) and investigate two operational paradigms: bi-level (decentralized) planning approach versus joint (collaborative) optimization. In the bi-level setting, each logistic service provider (LSP) plans its EV routes independently (deciding when/where to request charging), then a utility company dispatches MCSs to fulfill those requests. In the joint optimization, all EV routes (for multiple LSPs) and MCS routes are optimized simultaneously to minimize overall cost. We compare the two settings by conducting numerical experiments on small-size instances derived from a benchmark dataset. Results show that joint optimization consistently reduces total energy costs and often requires smaller fleets for serving customers, indicating that synchronized mobile charging through collaborative planning can significantly improve the efficiency of EV delivery operations.