<p>Technology convergence (TC), particularly multi-technology convergence (multi-TC), has emerged as a key driver of innovation in today’s rapidly evolving technological landscape. While multi-TC entails higher-order interactions among three or more technologies, most existing studies focus on pairwise combinations, overlooking the emergent synergies resulting from the simultaneous interaction of multiple technologies. To bridge this gap, we propose a novel framework that models multi-TC events as hyperedges in a temporal hypergraph, preserving their intrinsic higher-order interactions instead of simplifying them into dyadic approximations. To examine the temporal dynamics of multi-TC, we employ a cross-order correlation index to capture dependencies across different interaction orders. We further develop an enhanced CAT-Walk model that integrates semantic, structural, and temporal information for multi-TC prediction. Using patent data from the new energy vehicle sector covering 2001–2023, we find that multi-TC exhibits short-term temporal dependence, progressive transitions across interaction orders, and directional asymmetry in its evolutionary patterns. The prediction results show that the enhanced model achieves strong inductive forecasting performance. In addition, semantic representations derived from IPC explanatory texts outperform those derived from patent abstracts, indicating that taxonomy-aligned semantic information is more suitable for multi-TC prediction. This study advances TC research by shifting the focus toward a multi-TC perspective, offering a more realistic and predictive framework for understanding complex innovation dynamics.</p>