Energy consumption of mobile cellular communications is mainly due to base stations (BSs) that constitute radio access networks (RANs). 5G technologies are expected to improve the RAN energy efficiency while supporting the forecasted growth in data traffic. However, the evolution of the absolute RAN energy consumption with 5G deployment is not clear. Moreover, existing BS power models are mostly derived from legacy equipment. Therefore, this work presents a method to evaluate and to project the total energy consumption of broadband RANs. We use on-site up-to-date measurements to determine power models of 4G BSs, showing a linear relationship between power consumption and data traffic with a static traffic-independent power component. We then build a prospective power model of 5G BSs by scaling 4G models with respect to bandwidth, number of data streams, and expected technological improvements. We apply this method to the RANs in Belgium over the 2020–2025 period for six scenarios of 5G deployment. Results show that the static energy consumption accounts for a major part of the total RAN energy consumption, which implies that concurrently operating 4G and 5G RANs consumes more energy than using only one generation. The sleep mode feature of 5G technology can reduce its RAN static energy consumption, improving energy efficiency by 10 times compared to 4G. Finally, we estimate the absolute carbon footprint of 4G and 5G RANs by considering embodied and operating greenhouse gas emissions. They follow a clear upward trend for scenarios with extensive 5G deployment.