In response to the global educational transformation driven by generative AI, teachers’ AI-based Technological Pedagogical and Content Knowledge (AI-TPACK) has emerged as a key indicator of professional expertise. Taiwan’s advancements in semiconductor and AI technologies provide a distinctive context for integrating AI in education. This study aims to develop AI-TPACK competencies among preservice music teachers through a three-cycle curriculum tailored to Taiwan’s educational context, integrating the nation’s strong emphasis on digital education and teacher professionalism.The curriculum was implemented through an action research approach in the course “Teaching Materials and Methods in Music.” It addressed key challenges identified in previous courses, including limited proficiency in AI tools relevant to music instruction and a lack of opportunities to apply AI in authentic teaching contexts. The participants were 18 elementary school preservice teachers majoring in music. The program comprised three instructional cycles: the first developed AI Technological Knowledge (AI-TK) and AI Technological Content Knowledge (AI-TCK) through hands-on use of diverse AI tools in music-related activities; the second emphasized AI Technological Pedagogical Knowledge (AI-TPK) by guiding participants in designing and simulating AI-integrated lesson plans; the third provided practicum experiences in elementary classrooms, allowing participants to apply their AI-TPACK competencies in real-teaching settings.Findings suggest that a structured, phased approach can effectively cultivate AI-TPACK competencies among preservice music teachers. Data analysis revealed increased confidence and fluency in using AI tools, reflecting gains in AI-TK and AI-TCK. Evidence also indicated enhanced understanding of pedagogical applications, demonstrating progress in AI-TPK. While preservice teachers generally indicated that teaching in elementary classrooms enhanced their AI-related instructional competence, classroom observations revealed an overemphasis on the technical functions of AI tools. Their instruction tended to prioritize AI-TK over AI-TPK or AI-TCK, resulting in insufficient musical integration, limited aesthetic responsiveness, and weak adaptability to real-time classroom conditionsThese findings underscore the need for teacher education programs to promote more balanced, musically grounded, and context-responsive approaches to AI integration. Situated at the intersection of Taiwan’s technological innovation and culturally rooted music education, this study demonstrates how localized AI-integrated curricula can bridge global advances in educational technology with context-specific, culturally informed music teaching practices.