In semiconductor manufacturing, machines are usually qualified to process a limited number of recipes related to products. It is possible to qualify recipes on machines to better balance the workload on machines in a given toolset. However, all machines of a toolset do not have equal uptimes and may further suffer from scheduled and unscheduled downtimes. This may heavily impact an efficient recipe-to-machine qualification configuration. In this paper, we propose indicators for recipe-to-machine qualification management based on the overall toolset workload balance under capacity constraints. The models, deployed in industry, demonstrate that the toolset capacity must be considered while managing qualifications. Industrial experiments show how capacity consideration leads to an optimal qualification configuration and therefore capacity utilization.