The penetration of low-carbon technologies in power generation has challenged fossil-fuel-focused electric utilities. While the extant, predominantly qualitative, literature highlights diversification into renewables among possible adaptation strategies, comprehensive quantitative understanding of utilities’ portfolio decarbonization has been missing. This study bridges this gap, systematically quantifying the transitions of over 3,000 utilities worldwide from fossil-fuelled capacity to renewables over the past two decades. It applies a machine-learning-based clustering algorithm to a historical global asset-level dataset, distilling four macro-behaviours and sub-patterns within them. Three-quarters of the utilities did not expand their portfolios. Of the remaining companies, a handful grew coal ahead of other assets, while half favoured gas and the rest prioritized renewables growth. Strikingly, 60% of the renewables-prioritizing utilities had not ceased concurrently expanding their fossil-fuel portfolio, compared to 15% reducing it. These findings point to electricity system inertia and the utility-driven risk of carbon lock-in and asset stranding.

低碳技术在发电中的渗透已经挑战了以化石燃料为重点的电力公司。尽管现有的，主要是定性的文献强调了在可能的适应策略中向可再生能源的多样化发展，但对公用事业的资产组合脱碳的全面定量理解仍然缺失。这项研究弥合了这种差距，在过去的二十年中，系统地量化了全球超过3,000种公用事业从化石燃料发电到可再生能源的过渡。它将基于机器学习的聚类算法应用于历史性的全球资产级数据集，并提取其中的四个宏行为和子模式。四分之三的公用事业没有扩大其投资组合。在其余的公司中，有少数人在其他资产之前增长了煤炭，一半的人赞成天然气，其余的则优先考虑可再生能源的增长。令人惊讶的是，优先考虑可再生能源的公用事业公司没有停止同时扩大其化石燃料投资组合的比例，而减少这一比例的是15％。这些发现指向电力系统的惯性以及公用事业驱动的碳锁定和资产搁浅的风险。