The recrystallization and grain growth of a reactor-grade Zr-Nb-Sn alloy were experimentally investigated in the temperature range of 400–500 °C. Electron backscatter diffraction (EBSD) analyses were conducted to characterize grain growth and texture change caused by annealing. Hardness measurements were conducted to obtain the recrystallization fractions and were compared with the machine learning-based recrystallization analyses provided by AZtecCrystal. With good agreement with the hardness-based recrystallization fraction, the machine-learning-based method provides classification of deformed (recrystallized and grown) and undeformed (non-recrystallized) grains, thereby revealing the local behavior of dynamic grain growth. Recrystallization occurs in 4 h at 450 °C, with distinct grain growth after 24 h. Mechanisms of intergranular hydride connectivity decrease with grain growth are elucidated. The reduced hydride connectivity with grain growth primarily occurs by the formation of intra-granular hydrides and homogenization of grain boundary energies. The reduced hydride connectivity significantly increases cladding ductility. For cladding with a significant hydrogen content (i.e., ∼600 wppm), the increase in cladding ductility (i.e., strain energy density (SED)) owing to the reduced hydride connectivity supersedes the background matrix ductility enhancement associated with grain growth.