Molecular solar thermal (MOST) systems are working their way as a possible technology to store solar light and release it when necessary. Such systems could, in principle, constitute a solution to the energy storage problem characteristic of solar cells and are conceived, at a first instance, as simple molecular photoswitches. Nevertheless, the optimization of their different required properties is presently limiting their technological scale up. From the chemical perspective, we need to design a novel MOST system based on unconventional photoswitches. Here, by applying multi-configurational quantum chemistry methods, we unravel the potentialities of ad hoc-designed molecular photoswitches, which aim to photoproduce cubane or cubadiene as high-energy isomers that can be thermally (or eventually catalytically) reverted to the initial structure, releasing their stored energy. Specifically, while cubane can be photoproduced via different paths depending on the reactant tricycle diene conformation, an undesired bicyclic by-product limits its application to MOST systems. An evolution of this starting design toward cubadiene formation is therefore proposed, avoiding conformational equilibria and by-products, considerably red shifting the absorption to reach the visible portion of the solar spectrum and maintaining an estimated storage density that is expected to overcome the current MOST reference system (norbornadiene/quadricyclane), although consistently increasing the photoisomerization energy barrier.