In this paper, a generic optimization problem arising in supply chain design is modeled in a game theoretic framework and solved as a decentralized problem using a mechanism design approach. We show that the entities in a supply chain network can be naturally modeled as selfish, rational, and intelligent agents interested in maximizing certain payoffs. This enables us to define a supply chain design game and we show that the well known Groves mechanisms can be used to solve the underlying design optimization problem. We illustrate our approach with a representative three stage distribution process of a typical automotive supply chain.