The Future of Computing for Materials Science Challenges

Abstract

Materials discovery increasingly relies on the coordinated use of theory, computation, experiment, data-driven methods, and emerging quantum technologies, yet the full potential of these tools is realised only when they operate within workflows that reflect the complexity of real systems. This perspective summarises current capabilities, limitations, and opportunities across these domains, drawing on contributions from academia, industry, and national laboratories to identify the scientific and structural requirements for more reliable and efficient discovery. Classical simulations provide broad coverage across design spaces, while experimental measurements reveal degradation, heterogeneity, and kinetic processes that determine performance under realistic conditions. Machine learning accelerates exploration when supported by well-curated datasets with clear provenance and uncertainty quantification, and quantum computing offers promising routes into correlated electronic behaviour when aligned with properties that influence engineering decisions. Collectively, these insights highlight the need for reproducible workflows, shared data standards, realistic benchmarks, and a research culture that prepares scientists to work across paradigms. By integrating these methodological and organisational elements, the community can move toward discovery processes that deliver robust predictions, support confident decision making, and shorten the path from conceptual design to deployable materials.