We look at the long time behavior of potential Mean Field Games (briefly MFG) using some standard tools from weak KAM theory. We first show that the time-dependent minimization problem converges to an ergodic constant −λ, then we provide a class of examples where the value of the stationary MFG minimization problem is strictly greater than −λ. This will imply that the trajectories of the time-dependent MFG system do not converge to static equilibria.Potential MFG are those games whose MFG system can be derived as optimality condition of the following minimization problemwhere (m, w) verifies the Fokker Plank equation −∂ t m + m + divw = 0 with m(0) = m 0 and the coupling function F in the MFG system is the derivative with respect to the measure of F. These games have been largely studied (see Lasry and Lions [25] for existence results and, among others [9,4,28] for further properties) but, so far, not much is known regarding their long time behavior outside the assumption of monotonicity, where Cardaliaguet, Lasry, Lions and Porretta [10] proved the convergence to the ergodic system