Recent research showed that the rotational degree of freedom in stacking 2D materials yields great changes in the electronic properties. Here, we focus on an often overlooked question: are twisted geometries stable and what defines their rotational energy landscape? Our simulations show how epitaxy theory breaks down in these systems, and we explain the observed behavior in terms of an interplay between flexural phonons and the interlayer coupling, governed by the moir{\'e} superlattice. Our argument, applied to the well-studied MoS2/graphene system, rationalizes experimental results and could serve as guidance to design twistronic devices.