This paper aims to make sense of the ways in which synthetic data renews the engagement with issues of knowledgeability and cultural situatedness in machine vision environments. The implementation of synthetic data poses new challenges for a critique on machine learning regimes, notably in moments when data sets become decoupled of their cultural situatedness and rest on beliefs of a data universality. Furthermore, my aim is to nuance the understanding of synthetic data by pointing to the complex hybridity of the synthetic. I argue that through the emergence of synthetic data, new questions need to be directed to when and how synthetically generated data comes to be a placeholder in data sets (Chun 2021; Mulvin 2021). I thus ask in what ways data scarcity can indicate a socio-political relationship between the lack and saturation of data existing in the world. For this entry, I rely on two cases in point to discuss the cultural meaning of synthetic data through critical data studies and cultural theory.