A system for using digital twins for scalable, model-based machine predictive maintenance comprises a plurality of digital twins and a simulation platform. The plurality of digital twins correspond to plurality of remotely located physical machines. Each respective digital twin comprises: product nameplate data corresponding to a unique physical machine, one or more simulation models, and a database comprising run time log data collected from sensors associated with the unique physical machine. The simulation platform is configured to process simulation models corresponding to the plurality of digital twins using a plurality of multiprocessor computer systems.