AMERICAN MINERAL REGISTRY EV Battery Mineral Statistics: Chemistries, Benchmarks and Demand https://americanmineralregistry.com/research/ev-battery-mineral-chemistry-statistics AMR inspected the IEA chart capture and its offered CSV, preserving the original download preamble, units and category labels. The chart was last updated 5 May 2021. Its notes define the full-vehicle boundary, excluded steel and aluminium, 75 kWh NMC622 cathode assumption and graphite-based anode. The original report executive summary supplies the rounded six-times characterization. The category table transcribes all nine source material columns for both vehicles. AMR calculates each electric-minus-conventional difference, its share of the total increment, the electric-side included-category composition and ratios where the conventional denominator is positive. The totals, selected-material contribution and aggregate ratio are calculated from those same input cells. No source zero is changed into a missing value or a universal technology claim; no missing value is replaced with zero. Market context comes separately from the IEA commentary published 13 February 2026 and its chart notes. The 2025 data are estimates. EV chemistry shares retain capacity weighting, four-wheeler coverage, LFMP inclusion and stockpiling exclusions. BESS statements and application-deployment shares keep their own population labels. Inequalities remain inequalities. AMR’s deployment index assigns 2020 = 100 and applies the reported sixfold relationship to obtain 2025 = 600, without inferring an absolute deployment total or intermediate series. The downloadable source-cell ledger gives the primary URL, table or paragraph locator, value and calculation for every presented table cell. The standalone reproduction program reads the supplied CSVs and verifies the arithmetic, total boundaries, defined ratios and preserved market bounds. Reproducibility here means reproducing AMR’s calculations and transcriptions. It does not mean independently rebuilding the IEA’s underlying vehicle model or market-estimation system. Run python3 reproduce.py in the extracted bundle directory. Original evidence stays at producer URLs.