Fragment-based methods offer a promising way to defeat the nonlinear scaling of conventional quantumchemistry by means of distributed computing, and thereby extend electronic structure calculations tolarge molecular systems including enzymatic reactions and protein–ligand binding. At present, quantumchemistry calculations in those domains are limited to mixed quantum/classical (QM/MM) methodswhose convergence is very slow with respect to the size of the QM region. Typical QM system sizes(50–100 atoms) may yield the wrong sign for enthalpy changes, and selection of an appropriate QM region must be pinned to experimental data and often rests on error cancellation. This talk will describemethods that can be used to obtain converged results at the ab initio limit (including methods beyonddensity functional theory), in an affordable manner on research-group-scale hardware.