A data-driven approach to forecasting end-of-life photovoltaic module waste

Photovoltaic (PV) technology is central to the transition towards net-zero energy systems. Today’s deployment, however, translates into a growing wave of electronic waste once modules reach end of life. Robust projections of future waste streams are therefore critical for the timely planning of collection, transport, and treatment infrastructure, and for enabling recycling at scale, thereby reducing reliance on primary raw materials. We project global PV waste generation through 2050 using both Weibull-based and Gaussian failure models, extending these established approaches by incorporating context-specific module service lives derived from a large-scale meta-analytic synthesis of empirical evidence on PV module degradation, and by explicitly accounting for the repowering demand arising from module failures over time. Technology- and mounting-specific service lives are estimated under globally representative climatic conditions and applied across scenario-based deployment pathways ranging from a current-policies baseline to an ambitious 1.5°C-aligned net-zero transition. Annual waste in 2050 could range from 35.24 to 62.69 million tonnes (Mt), with cumulative waste through 2050 consequently spanning 501.88–752.76 Mt. The choice of failure model primarily redistributes waste over time, whereas deployment assumptions dominate cumulative burdens. Thin-film waste remains minor across all scenarios, staying below 2.8 Mt per year. Relative to earlier studies relying on uniform lifetime assumptions and old deployment projections, the context-specific parametrisation adopted here, grounded in empirical degradation data, yields substantially higher waste estimates, underscoring the importance of differentiated lifetime assumptions for planning PV waste management and end-of-life infrastructure.