R&D tax credit has become one of the main innovation policy instruments in Iran in recent years. However, despite the 100 percent rate stipulated in the Knowledge-Based Production Leap Law, the actual uptake of this instrument by large firms has remained below expectations. This study aims to explain the role of policy architecture in shaping the relationship between legal generosity and the actual use of the R&D tax credit by large firms. The research is applied in purpose and qualitative in method. Data were collected in 2025 through semi-structured interviews with 10 firms and 10 policymakers, implementers, and researchers in this field, while the findings cover the period from 2023 to 2025. Data were analyzed using Strauss and Corbin’s three-stage coding procedure. The findings show that the use of this instrument by large firms is determined not merely by the nominal credit rate, but also by actual R&D capacity, the predictability of policy implementation, administrative and compliance costs, and perceived organizational risk. Accordingly, the article introduces the concept of the “gap between nominal generosity and effective generosity” and argues that the effectiveness of this instrument depends less on the legal rate itself than on the quality of institutional design, clarity of rules, coordination among involved institutions, and implementation certainty. The policy implication is that improving the effectiveness of the R&D tax credit requires shifting policy attention from increasing the support rate to enhancing implementability, reducing uncertainty, and aligning the instrument with the organizational realities of large firms.