Abstract: To address the issue of more efficient utilization of computing resources across different time zones, this invention provides a global cross-time-zone computing power scheduling method. The computing platform performs the following steps: receive computing power demand information from at least one time zone and store it in a storage module; according to objective function and constraints, use the machine learning module, based on the machine learning algorithm, the heuristic algorithm, or deep learning algorithm module to calculate a computing power scheduling plan that maximizes the profitability and utilization rate of the computing platform; and, allocate GPU resources in at least one time zone according to the computing power scheduling plan.