Title: Endogenous learning for green hydrogen in a sector-coupled energy model for Europe

Abstract: To reach climate neutrality by 2050 the European energy system would have to be radically transformed. This raises the question of how much this transition is going to cost and which technologies should be supported to enable the transformation. Although many energy system models deal with cost-optimal pathways, they typically neglect either the dependency of investment costs on the cumulative deployed capacity of the technology, the temporal variability of renewable energy sources or they only depict individual sectors. In this study, we want to address these weak points. We use a European energy system model with full sector-coupling, renewable variability and with endogenous temporal-delayed learning on renewable generation technologies (solar PV, onshore and offshore wind) and hydrogen electrolysis. Three different CO$_2$ budgets corresponding to +1.5$^\text{o}$C, +1.7$^\text{o}$C, +2.0$^\text{o}$C temperature increase are applied. First we show that for a +1.5$^\text{o}$C budget scenario it is optimal to deploy hydrogen electrolysis at large scale already by 2030, which reduces the electrolysis investment cost to 75 EUR/kW$_{\text{elec}}$ by 2050. Second, we compare three different methods to represent cost reductions and show that modeling without endogenous learning leads to an overestimation of the total annual system costs, especially in the case of ambitious climate targets, of up to 17%.