In systems where digitization is prevalent, the plethora of data involved presents emerging challenges for data analysis and custodianship of data-driven AI/ML models. When the data involves time scales related to system dynamics, the dynamic data-driven application systems (DDDAS) paradigm offers a highly effective framework for modeling purposes. However, there are systems where tracking data and models generated by the data over long periods of time (possibly decades or centuries) may be required. An example of such a system is the nuclear fuel cycle whose digitization may provide flexibility and enhanced security. The problem may be compounded if the data is spread across secured but dispersed computational platforms (e.g., cloud vs edge computing) in the ever-evolving computational architectures of cyberspace. For such systems the use of AI/ML is a requirement, provided that the models produced are trustworthy and the evolution of such systems can be pivoted to an origin that one can always backtrack to. The mathematical structure of spirals may offer a possible augmentation of AI/ML and DDDAS ensuring tractability during the evolution of a system over decades or centuries, while maintaining the capacity to return to a reference origin. Illustrative examples demonstrate the features, boundaries, and possible limitations of the spiral framework.