Commercializing responsible artificial intelligence (RAI) involves translating ethical principles for developing, deploying, and using AI into business models. However, prior studies have reported tensions between commercial interests (e.g., development speed or accuracy) and societal interests (e.g., privacy or human rights) that can undermine RAI’s value proposition. Conceptually, we distinguish two business model development perspectives on AI and responsibility: innovating responsible business models leveraging AI and designing RAI business models. Taking the second perspective, we investigate the value proposition of RAI through business model design by employing a two-stage research approach consisting of focus groups and member checking. Empirically, we present the learnings from identifying the design elements for RAI business models. These include two themes that can underlie such business models: providing vs. enabling RAI systems and the observation that the tensions in RAI’s value proposition are paradoxical, not dilemmas. With our conceptual groundwork and empirical insights, we make three contributions that offer critical considerations for RAI business model design. First, we conceptualize two pathways for designing RAI business models: a corner path to commercialized RAI systems vs. direct path to commercialized RAI systems. We argue that these paths have distinct implications for the responsible in RAI. Second, we reflect the sociotechnical nature of RAI systems by emphasizing the criticality of the social for responsibility. Third, we outline a research agenda for developing RAI business models.