“Commentary: The Materials Project: A materials genome approach to accelerating materials innovation,” by Anubhav Jain, Shyue Ping Ong, Gerbrand Ceder, Kristin Persson and colleagues, was published in APL Materials in July 2013. It announced the Materials Project, an effort funded under the U.S. Materials Genome Initiative to pre-compute the properties of inorganic materials with quantum-mechanical calculations and make the results freely available online.
The idea is high-throughput computing: rather than measuring properties one material at a time in the lab, the project runs density functional theory calculations across tens of thousands of known and hypothetical compounds, then publishes the resulting energies, structures, electronic and other properties in an open database with tools for browsing and bulk data mining. Centered at Lawrence Berkeley National Laboratory, it has grown into one of the most heavily used resources in computational materials science.
The Materials Project matters to the AI story because that open, structured dataset became essential training and validation fuel for the machine-learning models that followed, including GNoME and MatterGen, and its companion calculations help check whether AI-proposed materials are actually stable. It is a reminder that the recent breakthroughs in AI for materials rest on years of patient, shared data infrastructure.