Computational tasks such as model checking, controller synthesis, reliability engineering and automated planning frequently rely on the construction of so-called state spaces (or search spaces). These state spaces are directed graphs, and they tend to be enormous for industrial-sized problems. In this project, we investigate how GPUs can be used to massively accelerate state space construction and analysis. We build on our earlier pioneering work to make the techniques applicable in practice. A new input language will make the specification of problems easier, and our focus on data compression and distributed computation will allow the exploring of very large state spaces.