The proliferation of distributed energy resources and the integration of multiple energy carriers are reshaping our energy system from a highly regulated, operator-controlled network into a sophisticated Electric Internet of Things. This leads to two-side unpredictability in both uncertain renewable sources and more random user behaviours, leaving a large number of idle resources and unmet loads simultaneously. Therefore, an innovative paradigm that can unlock the unexplored flexibility in the energy network is desired. One successful business model with a similar initiative is the sharing economy in other sectors. Inspired by that, I explore the potential of energy sharing by answering why, how, and what if. For why, three typical approaches, i.e., self-sufficiency, sharing, centralized operation, are compared under a stochastic framework, and sharing stands out for the fact that it can achieve a nearly social optimal efficiency in a distributed manner. For how, I design effective mechanisms for sharing among node-level and networked prosumers, respectively. Several desirable properties of the sharing market equilibrium are proved theoretically. I also develop a bidding process for implementation with proof of its convergence under mild conditions. For what if, a two-stage robust economic dispatch problem is used as an example by replacing the centralized real-time operation with a distributed sharing scheme. I prove that the flexibility of the system is retained with only a slight sacrifice of the economy. These works reveal fundamental understandings that facilitate in-depth analyses of the future energy sharing market.