Despite significant progress in wireless sensor network’s (WSN) applications in recent years, major problems persist, notably the efficiency of the battery power use for data exchange. In fact, sensor nodes are known to be energy-limited, they are powered by a relatively low capacity, non-rechargeable battery, and in most cases, the nodes are deployed in inhospitable or hard-to-reach areas and are unlikely to be recoverable. Therefore, any program running on a smart sensor must consider energy management. This paper describes, for the first time, a comprehensive p-median based mathematical model of the WSN-IoT network's clustering problem. A set of decision variables is defined in order to express the objective of minimizing the energy consumption as a mathematical function to be minimized taking into consideration the residual energy of the nodes, among other constraints to be satisfied. Unlike most of the techniques proposed in the literature, this one is remarkably notable by being scalable to large sized networks with up to 90,000 nodes. Moreover, the proposed approach allows the dynamic determination of the optimal number “p” of clusters to be formed and the assignment of the sensor nodes to each cluster to decrease and balance energy consumption of the network even when the base station is located outside the network. A considerable energy savings of nearly 20% is obtained by applying our new proposed method compared to the literature results. This approach will undoubtedly play a key role in future studies in the field of IoT.