We study the effect of ambiguity on corporate financial decisions using a non-Bayesian multiprior approach. We use two models of decision making under ambiguity, Minimum Expected Utility (MEU) due to Gilboa and Schmeidler (1989) and Consensus Expected Utility (CEU) due to Bewley (2002), to study a canonical corporate finance model of real investment with expansion and contraction options. We show that MEU decision makers behave as pessimistic single-prior decision makers: they are reluctant to invest and eager to abandon. On the other hand CEU decision makers act very differently: they too are reluctant to invest but reluctant to abandon projects (escalating commitment). Moreover, anticipation of future reluctance to abandon may induce CEU agents to forego investment in the first place. In this setting, we show how financial contracts, such as convertible bonds, can be used to make the initial investment attractive. We also argue that CEU is a reasonable model of group decision making within corporations and identify related empirical predictions.