The matterverse is vast and complex. It comprises the infinite combinations of elements of the periodic table in ordered and disordered arrangements. In this talk, I will discuss the development of matterverse.ai, a new database and machine learning (ML) prediction platform for materials properties based on graph deep learning. A complement to existing ab initio databases such as the Materials Project, matterverse.ai focuses on probing the matterverse at scales not possible with ab initio methods, for example, predicting properties for millions/billions of hypothetical materials, long-time-scale dynamic simulations, etc. In addition, matterverse.ai will serve as a platform for the sharing of containerized ML models for materials simulations and property predictions. Finally, I will share our vision and future priorities for matterverse.ai, such as leveraging on active learning loops with ab initio databases to continually enhance prediction performance, targeting high-value properties, etc.