This paper argues that we ought to conceive of the Dark Energy problem—the question of how to account for observational data, naturally interpreted as accelerated expansion of the universe—as a crisis of underdetermined pursuit-worthiness. Not only are the various approaches to the Dark Energy problem evidentially underdetermined; at present, no compelling reasons single out any of them as more likely to be true than the other. More vexingly for working scientists, none of the approaches stands out as uncontroversially preferable over its rivals in terms of its rationally warranted promise, i.e. the reasons to further work on, explore, and develop it. We demonstrate this claim by applying a Peircean economic model of pursuit-worthiness in terms of a cognitive cost/benefit estimate—with the instantiation of theory virtues as key indicators of cognitive gains—to the four main Dark Energy proposals (the cosmological constant approach, modified gravity, quintessence, and inhomogeneous cosmologies). According to our analysis, these approaches do not admit of an unambiguous, or uncontroversial, ranking with respect to which ansatz deserves distinguished attention and research efforts. The overall methodological counsel that our analysis underwrites recommends a pragmatic double research strategy forward: to encourage and foster theory pluralism and the search for tests—with the goal of enhancing the testability of the $$\Lambda $$CDM model and “testing it to destruction".