The smart grid systems, in replace of the traditional power grid systems, have been widely used among some well-known telecommunication, IT and power industries. These systems have multiple advantages such as energy efficiency, reliability and self-monitoring. To prevent power outage, threshold based power usage control (PUC) in a smart grid considers a situation where the utility company sets a threshold to control the total power usage of a neighborhood. If the total power usage exceeds the threshold, certain households need to reduce their power consumption. In PUC, the utility company needs to frequently collect power usage data from smart meters. It has been well documented that these power usage data can reveal a person’s daily activity and violate personal privacy. To avoid the privacy concern, privacy-preserving power usage control (P-PUC) protocols have been introduced. However, the existing P-PUC protocols are not very efficient and the computations cannot be outsourced to a cloud server. Thus, the utility company cannot take advantage of the cloud computing paradigm to potentially reduce its operational cost. The goal of this paper is to develop a P-PUC protocol whose computation/execution is outsourceable to a cloud. In addition, the proposed protocol is much more efficient than the existing P-PUC protocols. We will provide extensive empirical study to show the practicality of our proposed protocol.