<p>The Internet of Underwater Things (IoUT) utilizes Autonomous Underwater Vehicles (AUVs) to aid in data collection and mitigate the data transfer burden on underwater nodes. Nevertheless, determining the optimal traversal routes for multiple AUVs is crucial to further minimize the transmission load on these nodes. To address this, our study introduces an intelligent multi-AUV movement plan to reduce and balance transmission loads across underwater nodes, thus extending the lifetime of the network. The proposed strategy arranges the network into a tree structure, balances the resulting tree, and then uses a dynamic neighbor search process to determine the AUVs movement plan in the three-dimensional underwater environment. The simulation results show that the proposed balanced approach not only finds optimal paths for multiple AUVs, but it also significantly reduces node transmission loads and energy consumption compared to unbalanced approaches. Furthermore, the findings demonstrate the superiority of load-based path initialization over uniform-based, which directs AUVs to prioritize nodes with higher data loads. Additionally, our findings show that utilizing AUVs’ multiple speeds allows for thorough exploration of the solution space, improving the algorithm’s performance in the complex three-dimensional underwater environment.</p>