In semiconductor assembly and test manufacturing (ATM), a station normally consists of multiple machines (maybe of different types) for a certain operation step. It is critical to optimize the utilization of ATM stations for productivity improvement. In this paper, we first formulate the bottleneck station scheduling problem, and then apply ant colony optimization (ACO) to solve it metaheuristically. The ACO is a biological-inspired optimization mechanism. It incorporates each ant agent's feedback information to collaboratively search for the good solutions. We develop the ACO-based scheduling framework and provide the system parameter tuning strategy. The system implementation at an Intel chipset factory demonstrates a significant machine conversion reduction comparing to a traditional scheduling approach.