Abstract: Despite the recent advances in physical simulations and machine learning, the exploration of novel inorganic crystals remains constrained by the expensive trial-and-error approaches. Recent developments in deep learning have shown that models can showcase emergent predictive capabilities with increasing data and computation in fields such as language, vision, and biology. In this talk, I will present our recent results on how graph networks trained at scale can reach unprecedented levels of generalization, improving the efficiency of materials discovery by an order of magnitude. Building on the 48,000 stable crystals identified in ongoing studies, improved efficiency enables the discovery of 2.2 million stable structures with respect to the current convex hull, many of which had escaped prior human chemical intuition. The scale and diversity unlock surprising modeling capabilities for downstream applications, leading in particular to highly accurate and robust learned interatomic potentials that can be used in condensed-phase molecular dynamics simulations and high-fidelity zero-shot predictions.