EXTENDED ABSTRACT: Electrifying and decarbonizing key industrial sectors, including chemical and materials industries, as well as transportation and aviation, is a pressing mission of our time. Industrial players and governments worldwide have set ambitious targets to reduce carbon emissions, which need to be decreased to net-zero around mid-century in order to address global warming and climate change. For such deep decarbonization, it is crucial to develop unprecedented materials, such as electrocatalysts that can efffciently and steadily produce green chemicals and fuels at scale and generate renewable electricity on demand. In this talk, I will introduce how a joint computational–experimental approach can be established to accelerate catalyst materials design using ffrst-principles atomistic simulations, synchrotron X-ray spectroscopies, and physics-inspired machine learning. I will highlight how mechanistically elucidating and quantitatively controlling the electronic structure of transition metal compounds can effectively alter their chemical bondings, modulate key reaction barriers toward dissolution and electrocatalysis, and thus inffuence their durability and reactivity, respectively, offering physics-driven guiding principles for optimizing their catalytic performance. Moreover, I will demonstrate how combining machine learning with materials physics can accelerate the discovery of multicomponent oxides—a new class of materials with too-complicated atomic orderings for an exhaustive investigation—through the development of machine-learning models that accurately predict orderingdependent materials properties by capturing intrinsic symmetries. To conclude, I will discuss new opportunities for further resolving fundamental knowledge gaps between the structural, reaction, and compositional complexities of electrocatalyst surfaces to fully unlock transformative materials design for catalyzing sustainability and decarbonization.