Interference is a challenging problem faced by communication systems where multiple users share a common spectrum. When the spectrum is managed dynamically, this problem has been addressed using two approaches. First, a centralized solution based on the maximization of the system sum-rate subject to individual power constraints. Second, a distributed approach based on game theory, where each user maximizes its rate selfishly. In this talk, we analyze the maximum attainable system sum-rate obtained from the Nash solutions as the power budget is increased towards infinity. The analysis is based on an optimization problem formulation, in particular a MPCC (mathematical program with linear complementarity constraints), in which we seek to maximize the system sum-rate over the set of Nash equilibria. To examine the desired asymptotic behavior of the maximum system sum-rates, we introduce a homogenization of this problem and provide sufficient conditions for the maximum objective value of the homogenized problem to equal the limit of decentralized maximum sum-rates as the users' power budgets tend to infinity. We also characterize when such a limit is equal to infinity and provide a constructive test for this to hold. Finally, we present a simplified analysis for the case of two users, and a special case that rules out the presence of the Braess-type paradox.