We study a distributionally robust optimization formulation (i.e., a min-max game) for problems of nonparametric estimation: Gaussian process regression and, more generally, linear inverse problems. We choose the best mean-squared error predictor on an infinite-dimensional space against an adversary who chooses the worst-case model in a Wasserstein ball around an infinite-dimensional Gaussian model. The Wasserstein cost function is chosen to control features such as the degree of roughness of the sample paths that the adversary is allowed to inject. We show that the game has a well-defined value (i.e., strong duality holds in the sense that max-min equals minmax) and show existence of a unique Nash equilibrium that can be computed by a sequence of finite-dimensional approximations. Crucially, the worst-case distribution is itself Gaussian. We explore properties of the Nash equilibrium and the effects of hyperparameters through a set of numerical experiments, demonstrating the versatility of our modeling framework.