Extended Data Fig. 9: Multi-factor sensitivity analysis of 2050 cost estimates for 22 technologies.

All 22 cost drivers identified in Table S5 were simultaneously varied in the 1,000 cost combinations drawn by the Latin-hypercube sampling method from the input space (10⁴⁴ possible combinations). a) Box-and-whisker plots display the median (line), interquartile range (box), and 5th–95th percentiles (whiskers) for each technology, with violin plots representing the probability density of cost distributions. b) Statistical summary of a), reporting the relative standard deviation (RSD) and quartile coefficient of dispersion (QCD) for each technology. Both RSD and QCD values are shown as percentages in this figure.