A salient aspect of research in batteries is the discovery of new materials or novel properties of existing compounds. In this quest, the traditional approach is to focus on archetype compounds in which a desirable property was first observed, stimulating further investigations. This has been the innovation pathway of many battery materials (e.g., LiCoO₂, LiFePO₄) which are found in commercial Li-ion batteries. Indeed, most Li-ion cathode materials belong to a handful of structural families: layered oxides, spinel, and olivine. Similarly, many good Li-ion solid electrolytes crystalize in a few crystal structures: perovskites, NASICON-like, LISICON-like, garnets, and argyrodite. Yet this trial-and-error exploratory research based on extrapolating known solutions to new compositional spaces has high demands in terms of synthesis and characterization times that hinder the emergence of new, disruptive compounds.

In this contribution we will present the strategies, using high-throughput bond-valence calculations and machine learning, that have allowed us to develop new high-voltage materials for the Li-ion and Na-ion batteries. Following this approach our recent results related to the nitridophosphate family (Na₃TM(PO₃)₃N), where N assists an inductively induced high redox potential, will be shown. Within this family, Na₃V(PO₃)₃N has been found to be the Na-ion cathode material with the highest operation voltage for the V^IV+/V^III+ redox couple together with Na₇V₃(P₂O₇)₄. Strategies to further lower the cost and improve the safety of these systems will be shown.