Penalized splines are popular for nonparametric regression. We establish the minimax rate optimality of penalized splines for uniform convergence, thus improving the existing rate in the literature. The result is applicable to several types of penalized splines that are commonly used and holds under mild conditions on the design points.