Drastic variations of energy costs are witnessed in past decades, especially for low-carbon technologies, where decreasing and increasing trends co-existed. Estimating the cost evolution in the future is hence essential in long-term energy planning. Despite a number of existing studies, the estimated costs show strong hetero- geneity. Additionally, emerging technologies, such as electrolysis and CCUS (carbon capture, utilisation and storage), have gained limited attention. To improve the plausibility of the cost projection, we analysed the rela- tionship between accumulated installation and the corresponding CAPEX for 14 low-carbon technologies, and applied 5-8 learning curves (LRs) via Non-Linear Optimization (NLP) for projecting the cost evolutions towards 2050. The LRs were carefully selected based upon the index of Coefficient of Determination, and calibrated by comparison to a bunch of existing literature. Based upon our results: (1) residential PV and onshore wind rank the highest and lowest respectively in terms of the decreasing potential; (2) the majority of energy technologies are promising to achieve 36% - 74% cost reduction in 2050 compared to 2020, with a mean value around 50%. This study can be helpful as benchmark for energy stakeholders in decision-making towards carbon neutrality.