We investigate the metallicity evolution and metal content of the intergalactic medium (IGM) and galactic halo gas from z= 2 to 0 using 110‐million‐particle cosmological hydrodynamic simulations. We focus on the detectability and physical properties of ultraviolet resonance metal‐line absorbers observable with Hubble’s Cosmic Origins Spectrograph (COS). We confirm that galactic superwind outflows are required to enrich the IGM to observed levels down to z= 0 using three wind prescriptions contrasted to a no‐wind simulation. Our favoured momentum‐conserved wind prescription deposits metals closer to galaxies owing to its moderate energy input, while the more energetic constant wind model enriches the warm‐hot IGM 6.4 times more. Despite these significant differences, all wind models produce metal‐line statistics within a factor of 2 of existing observations. This is because , , and absorbers primarily arise from T < 105 K, photoionized gas that is enriched to similar levels in the three feedback schemes. absorbers trace the diffuse phase with , which is enriched to ∼1/50 Z⊙ at z= 0, although the absorbers themselves usually exceed 0.3 Z⊙ and arise from inhomogeneously distributed, unmixed winds. Turbulent broadening is required to match the observed equivalent width and column density statistics for . and absorbers trace primarily T∼ 104 K gas inside haloes (), although there appear to be too many absorbers relative to observations. We predict the COS will observe a population of photoionized absorbers tracing T < 105 K, gas with equivalent widths of 10–20 mÅ. and are rarely detected in COS signal‐to‐noise ratio 30 simulated sight‐lines (dn/dz≪ 1), although simulated detections trace halo gas at T= 106–107 K. In general, the IGM is enriched in an outside‐in manner, where wind‐blown metals released at higher redshift reach lower overdensities, resulting in higher ionization species tracing lower density, older metals. At z= 0, 90 per cent of baryons outside galaxies are enriched to , but 65 per cent of unbound baryons in the IGM have and contain only 4 per cent of all metals, a large decline from 20 per cent at z= 2, because metals from early winds often re‐accrete on to galaxies while later winds are less likely to escape their haloes. We emphasize that our results are sensitive to how metal mixing is treated in the simulations, and argue that the lack of mixing in our scheme may be the largest difference from other similar publications.