Infrastructure systems are increasingly shaped by interdependencies and emergent behaviors. These dynamics, intensified by climate change, urbanization, and technological advancement, demand rapid, holistic, and adaptive responses that traditional models often fail to provide (Mitchell, 2009). As illustrated by the 2025 California wildfire, such emergent phenomena—ranging from immediate impacts to secondary environmental risks— underscore the need for real-time, adaptive interventions (Entcheva, 2025; Mitchell, 2009; Peter and Swilling, 2014). While current DTs provide valuable insights within specific sectors (Tang et al., 2024), their isolated nature limits their capacity to reflect and manage broader systemwide behaviors. The growing body of literature highlights that digital twins must evolve to reflect these cross-domain interdependencies. Complexity science provides a valuable lens for understanding such systems, noting properties like self-organization, emergence, and adaptive feedback loops. By treating multiple digital twins as a single complex ecosystem, the paper discusses the theoretical underpinnings of Complex Digital Twins (CoDTs), defines their structure and variants, and illustrates their potential through real-world applications, aiming to guide future research, development, and governance in complex infrastructure environments.