The study presents an approach for searching optimal borehole placement for shallow foundation design under spatially variable conditions. A recently proposed approach named Random Failure Mechanism Method is adopted, which allows for 3D bearing capacity estimations considering spatially variable soil and given borehole locations. The borehole placement problem is formulated as finding the borehole locations that minimize the standard deviation of the foundation bearing capacity. A stochastic optimization framework named Asymptotic Bayesian Optimization is implemented to handle the inherent variability of the standard deviation estimates. The applicability of the proposed approach is demonstrated based on a scenario involving a rectangular footing and two boreholes. The feasibility and effectiveness of the approach are promising for future applications, mostly for proposing optimal borehole placement for typical engineering practice foundation layouts.