There is a growing consensus that we must look beyond IQ to understand the mechanisms of talent development. Grounded in the Actiotope Model of Giftedness, this study adopts a resource-based approach and examines the incremental and interactive contributions of educational and learning capital to STEM achievement beyond IQ. Data were collected from 318 German secondary school students (grades 6–10; Mage = 12.08; 50.3% male) using domain-specific measures of educational and learning capital, a nonverbal matrix intelligence test, and STEM grades. Robust regression and mediation analyses showed that learning capital significantly predicted STEM achievement beyond general intelligence, whereas educational capital exerted no direct effect. Instead, the relationship between educational capital and achievement was fully mediated by learning capital. Moreover, the interaction term of educational and learning capital predicted achievement. A further interaction indicated that the positive effect of learning capital on STEM achievement was stronger for students with higher intelligence, consistent with an intelligence utilization (Matthew) effect. These findings support a systemic interpretation of achievement in which intelligence reflects prior resource utilization and functions as a catalyst, while current learning resources constitute the proximal determinants of STEM performance.