A system and method for dynamic geospatially-referenced cyber-physical infrastructure inventory and asset management using state models, wherein a computing device with a geolocation device and wireless networking capability is attached to each of a plurality of physical assets, and used to periodically determine a state of the physical asset to which it is attached using the geolocation device, periodically generate a status update message and send it to a remote computer, and wherein the remote computer stores the status of the physical asset as time series data in a state model and, if a status message is not received in a defined period of time, applies a machine learning algorithm to the state model to predict a current or future state of that particular physical asset.