We propose a strategic communication model, in which social agents play a central role in determining how much effort and time they want to invest in interacting with others. The starting point for this paper is the so called bounded confidence model in which agents update their opinions only when they are like-minded (i.e., their opinion distance is smaller than a threshold). In our model, in addition to the existence of trust between interacting agents, the agents also individually determine the expected time and effort in interacting with their neighboring agents. The strategic communication thus refers to the process that allows individuals to select neighbors, with whom interaction produces a maximization of the local utility functions. Our goal is to analyze the dynamics of opinion formation under the proposed communication strategy, with a focus on understanding how and under what conditions clustering patterns emerge in opinion space.