Smart cities constitute a new urban paradigm and a hegemonic phenomenon in contemporary city development. The concept envisages a data-enhanced future and efficiency gains made possible by automation and innovation in city activities and utilities. However, the way smart cities are created brings about two weaknesses. First, there is strong compartmentation of solutions and systems, which are developing in vertical markets for energy, transport, governance, safety, etc., silos with little interoperability and sharing of resources. Second, there is a low impact, some increase in efficiency, some reduction in costs, time gained, some decrease in CO₂ emissions. There is an important knowledge gap about developing cross-sector, high-impact smart city systems. This paper deals with these challenges and investigates a different direction in smart city design and efficiency. We focus on ‘Connected Intelligence Spaces’ created in smart city ecosystems, which (a) have physical, social, and digital dimensions; (b) work as systems of innovation enabling synergies between human, machine, and collective intelligence; and (c) improve efficiency and performance by innovating rather than optimizing city routines. The research hypothesis we assess is about a universal architecture of high impact smart city projects, due to underlying connected intelligence spaces and cyber-physical-social systems of innovation. We assess this hypothesis with empirical evidence from case studies related to smart city projects dealing with safety (Vision-Zero), transportation (MaaS), and energy (positive energy districts). We highlight the main elements of operation and how high efficiency is achieved across these verticals. We identify commonalities, common innovation functions, and associations between functions, allowing us to define a common architecture enabling innovation and high performance across smart city ecosystems.