Architecture-level evaluations of Palladio currently lack support for the analysis of the power efficiency of software systems and the effect of power management techniques on other quality characteristics. This neglects that the power consumption of software systems constitutes a substantial proportion of their total cost of ownership. Currently, reasoning on the influence of design decisions on power consumption and making trade-off decisions with other Quality of Service (QoS) characteristics is deferred until a system is in operation. Reasoning approaches that evaluate a system’s energy efficiency have not reached a suitable abstraction for architecture-level analyses. Palladio and its extension SimuLizar for self-adaptive systems lack support for specifying and reasoning on power efficiency under changing user load. In this paper, we (i) show our ideas on how power efficiency and trade-off decisions with other QoS characteristics can be evaluated for static and self-adaptive systems and (ii) propose additions to the Palladio Component Model (PCM) taking into account the power provisioning infrastructure and constraints.