In the past few years, topological deep learning (TDL), a term coined by us in 2017, has become an emerging paradigm in data science and a frontier for relational learning. By leveraging algebraic topology’s high-level abstractions, TDL reduces dimensionality, simplifies geometric complexity, captures high-order interactions, and provides an interpretable framework. TDL models have been widely applied across science, engineering, medicine, defense, and industry. Notable demonstrations of TDL’s superiority include our team’s victories in the D3R Grand Challenges (a global competition series in computer-aided drug design), our discovery of SARS-CoV-2 evolutionary mechanism, and our successful prediction of dominant viral variants two months in advance.