From minerals to dollars: A scrutiny of lithium-ion battery cost models

The accelerating growth of electric mobility and stationary storage has intensified demand for lithium-ion batteries, where raw materials represent 50–70% of total cost. Volatility in Ni, Co and Li markets therefore poses major risks for affordability and investment decisions. This study systematically evaluates existing battery-cost models with a specific focus on how raw-material price inputs are represented, standardized and propagated into final $/kWh outputs. A comparative review of sixteen representative models, including BatPaC, GREET®, EverBatt and CellEST, investigates methodological assumptions, cost-element structure, and transparency of data sources. This work provides the first review that jointly analyses deterministic cost models alongside probabilistic validation using real-world commodity volatility. Robustness is assessed through a 1,000-run Monte Carlo simulation based on historical price distributions of key metals. Results indicate clear differentiation among chemistries. NMC-111 exhibits the widest 5–95% uncertainty range due to cobalt sensitivity, whereas NMC-811 and NMC-9525 show narrower bands reflecting reduced cobalt exposure. Bottom-up process-based models offer stronger cost traceability and better uncertainty representation than top-down macro forecasts, which often lack explicit linkages to $/kWh outputs. Findings demonstrate that raw-material price treatment strongly influences forecast accuracy, and that scenario-free point estimates systematically under-represent market risk. To address these gaps, the study proposes a structured multi-source data acquisition strategy integrating public databases, commodity-market feeds and industry publications. The review enhances transparency, establishes reproducible comparison criteria, and quantifies price-volatility effects, offering actionable guidance for researchers, analysts and policymakers aiming to improve reliability and reduce uncertainty in LIB cost assessment.