Cities worldwide have been implementing various strategies to mitigate congestion, pollution, and noise from urban last-mile logistics. However, most initiatives fail economically or are ignored by logistics service providers (LSPs), mainly because their perspectives are rarely considered. Creating sustainable last-mile logistics systems requires considering the often-conflicting objectives of all stakeholders. We address this challenge by investigating a novel decision-making problem where municipalities aim to repurpose existing, underutilized public spaces into shared microhubs, where LSPs can transfer goods from conventional freight trucks to sustainable vehicles for last-mile delivery. We conceptualize the problem as a stochastic single-leader, multi-follower Stackelberg game, where the municipality (i.e., the leader) aims to find the location of shared microhubs by explicitly accounting for LSPs' reaction and the stochasticity of the demand for the services they offer, while adhering to public budget constraints. We formulate the problem as a two-stage stochastic bilevel program. Since microhubs can be used by multiple LSPs, the lower-level problem constitutes a generalized Nash equilibrium problem due to the shared capacity constraints. We devise several structural properties to prove the existence of equilibria and characterize optimal lower-level decisions, which allow us to propose an efficient branch-and-Benders-cut approach to solve the problem.