Vehicle automation has already become commercially available. Today’s mainstream vehicles are equipped with Adaptive Cruise Control (ACC) to automate longitudinal car-following. However, ACC’s limited detection range and delayed reaction hinder the vehicle’s ability to respond promptly to speed changes, and ACC could amplify minor disturbance into severe stop-and-go waves and form queues at bottlenecks. Consequently, the discharge flow could be lower than the maximum flow observed in absence of queues or congested conditions, also known as capacity drop. Microscopic simulation of freeway bottlenecks consisting of a single lane with reduced speed zone, multilane with an on-ramp merge, and multilane with an off-ramp diverge demonstrates that ACC vehicles lead to capacity drop at freeway bottlenecks. For the single lane reduced speed zone, the extent of capacity drop for ACC could be either less or more severe than that of human driven vehicles. However, the multilane freeway on-ramp and off-ramp bottlenecks are more susceptible to capacity drop than human driven vehicles, due to ACC significantly amplifying disturbances caused by lane changes, merging, and diverging.