We explore a parallel and scalable reformulation of the guaranteed quality, multi-material, conforming, tetrahedral mesh generation algorithm known as lattice cleaving. The local scope of the algorithm's operations makes the technique highly amenable to parallelization. The main challenge in reformulating rests on decomposing the background mesh domain into suitable subdomains for each algorithmic phase. Such a reformulation of lattice cleaving enables a streaming, out-of-core solution for memory intensive datasets on architectures with limited resources.