Sung Eun Jerng
Solid-state batteries (SSBs) promise the step-change in energy density and safety that today’s lithium-ion technology cannot deliver, yet progress is throttled by the slow and fragmented discovery–manufacture–testing cycle for solid electrolytes. This review traces how machine learning (ML) is beginning to fuse those conventionally isolated stages into a coherent, data-driven pipeline. We first examine materials discovery workflows that screen millions of hypothetical electrolytes, screening into shortlists with high room-temperature conductivities and wide electrochemical windows. We then survey ML-accelerated fabrication research that routinely cut experimental iterations by 70–80% while optimizing the fabrication process. At the cell level, physics-informed neural networks and surrogate phase-field models are shown to predict cell failures with cycle-level fidelity, pointing to data-centric lifetime design rules. Finally, we highlight emergent self-driving laboratories in which robotics, high-throughput characterization, and active-learning planners run 24/7 closed-loop campaigns, already delivering previously unreported electrolytes in days rather than months. By stitching together advances across discovery, processing, and lifetime prediction, we argue that ML is poised to transition from a posthoc analysis tool to a real-time copilot that steers SSB development toward commercially viable, high-energy systems at unprecedented speed and scale.