A well-known characteristic of working memory (WM) is its limited capacity. The source of such limitations, however, is a continued point of debate. Developmental research is positioned to address this debate by jointly identifying the source(s) of limitations and the mechanism(s) underlying capacity increases. Here we provide a cross-domain survey of studies and theories of WM capacity development, which reveals a complex picture: dozens of studies from 50 papers show nearly universal increases in capacity estimates with age, but marked variation across studies, tasks, and domains. We argue that the full pattern of performance cannot be captured through traditional approaches emphasizing single causes, or even multiple separable causes, underlying capacity development. Rather, we consider WM capacity as a dynamic process that emerges from a unified cognitive system flexibly adapting to the context and demands of each task. We conclude by enumerating specific challenges for researchers and theorists that will need to be met in order to move our understanding forward.