Ocean vertical velocity (w) plays a key role in regulating the exchanges of mass, heat and nutrients between the surface and deep ocean. However, direct observation remains difficult due to its small magnitude and large spatiotemporal variability. Therefore, w fields are generally diagnosed using dynamic‐based methods. In this study, we developed a deep neural network (DNN) to reconstruct three‐dimensional fields of ocean vertical velocity based on sea surface height (SSH) fields. Compared to dynamic‐based methods, the DNN shows improved performance in the w reconstruction within upper 500 m in terms of higher correlation and less error. Remarkably, the DNN requires only a ∼45 × 45 km size SSH image as input to estimate w at the center. This suggests that the DNN has great potential for w reconstruction in the future combined with high‐resolution observations such as the Surface Water and Ocean Topography mission. Plain Language Summary: Direct measurement of ocean vertical motion is challenging because it is too weak and unstable. However, understanding this vertical motion is important to study the heat/material transports in the oceans and even climate variability. Here, we propose a deep learning‐based method that uses only the sea surface height (SSH) fields to reconstruct ocean vertical velocity. Our results indicate that this new reconstruction method outperforms the conventional method, particularly in the upper ocean. Moreover, the less data input makes it possible to be an effective tool for reconstructing oceanic vertical velocity with high‐resolution SSH observations in the future. This demonstrates the promise of using deep learning to estimate hard‐to‐measure ocean features, improving our knowledge of ocean processes related to climate. Key Points: A deep learning (DL) based approach is developed to reconstruct the ocean vertical velocityUsing the modeled sea surface height (SSH) as input, the DL approach outperforms the dynamic‐based methodThe DL approach has the potential to improve the retrieval of real‐time vertical velocity field along the SWOT swaths [ABSTRACT FROM AUTHOR]