The hydrostatic journal bearing equipped with a carbon-fiber-reinforced carbon-based porous bushing is employed in the hydraulic turbomachine. The bearing exhibits high load capacity, but may unduly consume pressurized lubricant. This study aims to maximize the load capacity and minimize the feeding power. The journal radius, nominal clearance, porous bushing length, porous bushing thickness, feeding pressure, and material permeability are selected to optimize. A fast optimization method is proposed, integrating an in-house porous journal bearing solver (PBS), sampling method, surrogate model, and genetic algorithm. Behind PBS, a theoretical flow model based on the Reynolds lubrication equation and the Darcy equation is established, and a new numerical method based on the finite difference method is proposed. PBS substitutes ansysfluent by calculating bearing performances accurately and instantly, which is the first novelty to facilitate optimization. Then, artificial neural networks are trained as error-free and time-efficient surrogate models to produce bearing objectives in the evolution, which is the second acceleration highlight. The running time is reduced significantly. The load capacity is improved by 68.1%, whereas the feeding power declines by 50.5%. In the optimized case, a sharp pressure hump leads to greater load capacity, while the radial velocity decreases, resulting in reduced feeding power.