<p>The loss of submarine cables is influenced by various factors such as human factors and the marine environment, which affect the transmission efficiency of electrical energy and the service life of cables. Studying the loss of submarine cables and predicting the loss has research significance. This paper investigates the prediction of submarine cable losses based on an improved slime mold algorithm. Two-dimensional models of submarine cables under three laying methods, including direct laying, buried laying, and pipeline laying, are established using COMSOL finite element simulation software. The losses of the cable core, shielding layer, and armor layer are calculated and analyzed when the cable reached its ampacity under different conditions, including burial depth, seawater temperature, transmission frequency, and seawater flow velocity. To address the problems of complex modeling, time-consuming calculations, and poor flexibility in simulation software, a multi-strategy improved slime mold algorithm based on Latin hypercube sampling, ReLU function, and Gaussian–Cauchy mutation is designed for loss prediction. This method can quickly predict the losses of specific submarine cables, which helps to select design and laying methods under different environments and usage requirements, maximizing energy transmission efficiency and reducing dependence on simulation software.</p>