Artificial Intelligence (AI) has emerged as a technology that can be applied in many areas, with the potential of revolutionizing the productivity, decision-making and competitiveness of Micro, Small and Medium Enterprises (MSMEs). However, the adoption of AI by MSMEs has been found to be less than by large enterprises across the globe and India and this is not just a technological, but an organizational and behavioral challenge. This paper proposes a conceptual framework — the Capability–Adoption Gap Model — based on the Technology–Organization–Environment (TOE) framework, Diffusion of Innovation (DOI) theory, and owner-manager behavioural-readiness constructs to understand the rationale behind the under-adoption of capable AI tools in the MSME ecosystem in Gujarat. The paper presents a narrative-synthesis of recent global, Indian and Gujarat-specific literature, and proposes four interacting sets of capabilities, namely technological, organizational, environmental/institutional and behavioural, that collectively influence adoption outcomes. Seven testable propositions are advanced that connect these dimensions to intention to adopt and actual use and implications for managerial and policy action for bridging the divide are presented. The framework provides researchers with a foundation for empirical testing and support agencies for MSMEs with a diagnosis tool to design interventions.