Large bone fractures often require porous implants for complete healing. In this work, we numerically investigate the suitability of three topologically very different architected materials for long bone implants: the octet truss-based lattice, the Schwartz P minimal surface-based lattice and the spinodal stochastic surface-based lattice. Each implant topology (reinforcement) and its surrounding tissue (soft matrix) are modeled as a composite system via finite element analysis. Performance metrics are defined based on the Young’s modulus, the peak stress under service conditions, the interfacial surface area per unit volume and the relative bone growth rate (estimated based on the strain transferred to the soft matrix). We show that surface-based topologies are less prone to fatigue failure and may promote supe- rior bone growth than conventional truss-based designs. Spinodal surface-based architected materials have the best performance, and can be fabricated via self-assembly approaches followed by material con- version, potentially allowing scalable fabrication of implants with unit cell sizes at the micro-scale, thus dramatically amplifying surface area per unit volume and bone growth efficiency.