Today, many e-government projects are of large scale. They can be considered as socio-technical ecosystems in which citizens, authorities, and enterprises collaborate through information systems. Several of these e-government projects are characterized as partial or total failures, because they could not master the high complexity resulting from a large number of actor classes, nontransparent collaboration processes, and heterogeneous IT landscapes. In an intra-organizational context, architectural thinking supports decision-making by providing continuous transparency on social and technical elements and their relations. This paper extends architectural thinking to the ecosystem level and positions it as an approach that both scholars and project managers can use to deal with the increasing complexity of e-government projects. By conducting a multiple case study of e-government projects, we identify seven areas of architectural concerns and develop a corresponding ecosystem architecture meta-model as the first steps towards leveraging architectural thinking for e-government projects.