The properties of polymer–nanoparticle (NP) mixtures significantly depend on the dispersion of the NPs. Using molecular dynamics simulations, we demonstrate that, in the presence of polymer–NP attraction, the dispersion of NPs in semidilute and concentrated polymers can be stabilized by increasing the polymer concentration. A lower polymer concentration facilitates the aggregation of NPs bridged by polymer chains, as well as a further increase of the polymer–NP attraction. Evaluating the binding of NPs through shared polymer segments in an adsorption blob, we derive a linear relationship between the polymer concentration and the polymer–NP attraction at the phase boundary between dispersed and aggregated NPs. Our theoretical findings are directly relevant for understanding and controlling many self-assembly processes that use either dispersion or aggregation of NPs to yield the desired materials.