Coupled thermo-mechanical part-scale simulations in laser powder bed fusion (LPBF) enable the prediction of residual stresses and thermal distortion without physical prototyping. While many existing simulation approaches rely on heuristic models to capture the effect of the heat source—thereby limiting their generalizability—scan-resolved simulations based on first principles can be applied across diverse processing conditions. This allows for the assessment of various influencing factors on the quantities of interest and the exploration of novel process strategies without costly experimental trial and error. However, these high-fidelity simulations are computationally expensive, requiring millions of time steps to accurately capture the multiscale physics of LPBF. To alleviate this burden, the inherent spatial and temporal multiscale nature of the process can be leveraged. Previous work has utilized a-priori mesh coarsening strategies motivated by process knowledge, as well as error-estimator-based adaptive mesh refinement, to adjust the mesh size to the physical length scales. The temporal multiscale nature of the thermal problem has been leveraged via multirate time integration, where degrees of freedom are classified as fast- or slow-evolving. For the mechanical subproblem, initial approaches split the solution process into a local subproblem—directly affected by the heat source and solved with a finer temporal resolution using boundary conditions derived from the remaining domain—and a global correction step where the entire domain is solved. The talk presents first steps towards the combination of such multiscale approaches for the coupled thermo-mechanical problem with high-performance computing. In particular, computational aspects for the efficient implementation of multirate time integration will be highlighted. Eventually, the efficiency potential of these methods will be explored to increase system sizes accessible by thermo-mechanical LPBF simulations.