In this study, a multiscale material model is employed to simulate two metal forming processes: 2D plane strain compression and a 3D biaxial bulge test. A generalized Taylor-type polycrystal model is employed to describe the fine scale viscoplastic response of the material, while the coarse scale response is computed using a multiphysics finite element code. The coupling between the local responses of the textured polycrystal and the continuum level is achieved via an adaptive sampling framework, which is shown to greatly reduce the total number of fine scale evaluations required to achieve a specified error tolerance. The anisotropy represented at the fine scale is sufficient to observe strain localization in both forming processes. For the case of idealized plane strain compression, a fairly diffuse yet distinct patterning of plastic strain localization develops in a manner consistent with experimental observations. The application of friction constraints to the compression surfaces—as is present in channel die compression tests—dramatically strengthens and redistributes the localization patterns. The simulated biaxial bulge test also demonstrates strain localization that is in agreement with the locations of diffuse necks in experimental observations. The tests are conducted using a federated multiple-program multiple-data simulation, which allows for load balancing between the coarse and fine scale calculations. Such a simulation framework is capable of efficiently embedding physically robust, but computationally expensive material models in component scale simulations appropriate to design decisions.