Approximately half of HIV patients continue to experience HIV-associated neurocognitive disorders (HAND). Our study aims to evaluate changes in the dynamic activity patterns of functional brain communities in the early stages of HIV infection by comparing time-varying multilayer network metrics. A total of 165 persons living with HIV but without neurocognitive disorders (PWND), 173 individuals with asymptomatic neurocognitive impairment (ANI), and 100 matched healthy controls (HC) were enrolled. A time-varying multilayer network model was constructed, and global modularity (Q value) and nodal flexibility were calculated using different parameter settings (γ = [0.9, 1, 1.1], ω = [0.5, 0.75, 1]). Brain functional alterations in the PWND and ANI groups were evaluated from both global and nodal perspectives. Associations between network measures, clinical variables, and cognitive performance were also explored. Using the full connectivity matrix, no significant differences in global modularity (Q value) were found among the three groups. However, when thresholding the matrix to retain the top 10% of strongest connections, the ANI group showed significantly lower modularity than the HC group across all γ and ω combinations (p < 0.05). At γ = 0.9 and ω = 0.5, reduced nodal flexibility was observed in visual network regions in the PWND group, while the ANI group showed reduced flexibility in regions belonging to the default mode network (DMN), sensorimotor network (SMN), and limbic network (LIM). At γ = 0.9 and ω = 1, the ANI group exhibited increased flexibility in DMN regions compared to HC. Additionally, thresholding the top 10% connections revealed increased flexibility in the right lingual gyrus (visual network) in ANI compared to HC (FDR corrected, p < 0.05). Nodal flexibility was positively correlated with neurocognitive performance in the PWND group, whereas a significant negative correlation was observed in the ANI group. Regardless of cognitive impairment, HIV patients exhibit abnormalities in dynamic community structures. These findings provide new insights and perspectives for the early detection of brain damage, advancing our current understanding of time-varying multilayer networks in HIV patients.