<p>This paper focuses on the asymptotic synchronization of mixed-delay neural networks with semi-Markovian jumping. At first, a stochastic sampled-data strategy concerning communication delay is devised by means of multiple stochastic sampling intervals. Next, by introducing multiple random variables to redefine the input delays and sampling interval, respectively, a synchronization error system is reconstructed. Subsequently, an innovative Lyapunov-Krasovskii functional (LKF) is established, which encompasses sampling points along with a cross term related to the communication delays and sampling intervals. Moreover, a less conservative asymptotic stability criterion in the mean square sense for the error system is derived using looped-functional terms. In the end, a simulation example is presented to showcase the effectiveness of the proposed synchronization criterion.</p>