A decentralized load management and control system and method are provided herein. A plurality of loads in a system are associated with a multitude of respective networked load controllers. If energy demand exceeds optimum limits in operation then the networked load controllers collaborate to determine which load(s) will be shed based on an optimization algorithm that considers, inter alia, variable load priority and business objectives. Additionally and/or alternatively, if the metered energy demand is less than optimum, the load controllers can determine which loads to reconnect again based at least upon load priority and business objectives.