We consider the problem of characterizing the locational marginal value of energy storage capacity in electric power networks with stochastic renewable supply and demand. The perspective taken is that of a system operator, whose objective is to minimize the expected cost of firm supply required to balance a stochastic net-demand process over a finite horizon, subject to transmission and energy storage constraints. The value of energy storage capacity is defined in terms of the optimal value of the corresponding constrained stochastic control problem. It is shown to be concave and non-decreasing in the vector of location-dependent storage capacities -implying that the greatest marginal value of storage is derived from initial investments in storage capacities. We also provide -as part of our main result -a characterization of said marginal value, which reveals its explicit dependency on a specific measure of nodal price variation. More generally, we derive an upper bound on the locational marginal value of energy storage capacity in terms of the total variation of the corresponding nodal price process, and provide conditions under which this bound is tight.