Direct epitaxial growth of III-V lasers on silicon provides the most economically favorable means of photonic integration but has traditionally been hindered by poor material quality. Relative to commercialized heterogeneous integration schemes, epitaxial growth reduces complexity and increases scalability by moving to 300 mm wafer diameters. The challenges associated with the crystalline mismatch between III-Vs and Si can be overcome through optimized buffer layers including thermal cyclic annealing and metamorphic layers, which we have utilized to achieve dislocation densities < 7×106 cm-2. By combining low defect densities with defect-tolerant quantum dot active regions, native substrate performance levels can be achieved. Narrow ridge devices with threshold current densities as low as ~130 A/cm2 have been demonstrated with virtually degradation free operation at 35°C over 11,000 h of continuous aging at twice the initial threshold current density (extrapolated time-to-failure >10,000,000 h). At 60°C, lasers with extrapolated time-to-failure >50,000 h have been demonstrated for >4,000 h of continuous aging. Lasers have also been investigated for their performance under optical feedback and showed no evidence of coherence collapse at back-reflection levels of 100% (minus 10% tap for measurement) due to the ultralow linewidth enhancement factor (αH < 0.2) and high damping of the optimized quantum dot active region.