Complex international partnerships have emerged as a policy instrument of choice for many governments to build domestic capacity in science, technology and innovation with the help of foreign partners. At present, these flagship initiatives tend to be primarily practitioner-driven with limited systematic understanding of available design options and trade-offs. Here, we present an analysis of four such partnerships from the university sector between the Massachusetts Institute of Technology (MIT) and governments in the UK, Portugal, Abu Dhabi, and Singapore. Using a system architecture approach in conjunctions with in-depth case studies and elements of interpretive policy analysis, we map how in each country distinct capacity-building goals, activities, and political and institutional contexts translate into different partnership architectures: a bilateral hub-&-spokes architecture (UK), a consortium architecture (Portugal), an institution-building architecture (Abu Dhabi), and a functional expansion architecture (Singapore). Despite these differences in emergent macro-architectures, we show that each partnership draws on an identical, limited set of ‘forms’ that can by organized around four architectural views (education, research, innovation & entrepreneurship, institution-building) and four levels of interaction between partners (people, programs/projects, objects, organization/process). Based on our analysis, we derive a design matrix that can help guide the development future partnerships through a systematic understanding of available design choices. Our research underscores the utility and flexibility of complex international partnerships as systemic policy instruments. It suggests a greater role for global research universities in capacity-building and international development, and emphasizes the potential of targeted cross-border funding. Our research also demonstrates the analytic power of system architecture for policy analysis and design. We argue that architectural thinking provides a useful stepping stone for STS-type interpretive policy analysis into national innovation initiatives in different political cultures, as well as more custom-tailored approaches to program evaluation.