Dynamic resource allocation in a cloud environment has become possible using virtualization technologies in cloud computing. One of the applications of these technologies is offering various applications by Software-as-a-Service (SaaS) infrastructures. Unfortunately, due to request rate increments in cloud rush hours, the related server cannot serve all the requests according to the service level agreement. Hence, the cloud provider’s quality of service will decrease. Thus a mechanism is required to control the admission rate of requests for cloud servers. In this study, an intelligent controller is designed and implemented on a field-programmable gate array (FPGA) in order to control the admission rate of requests for a SaaS server in the cloud. The controller is based on a brain emotional learning-based intelligent controller (BELBIC). First, an analytical model of a server is proposed and simulated, which shows the behavioural characteristics of a real server. Next, the BELBIC is designed to control the admission rate of the server. Finally, the system is implemented on FPGA hardware and simulated by a synthetic cloud workload in a hardware-in-the-loop manner. In order to compare the performance of the BELBIC, an adaptive neuro-fuzzy inference system (ANFIS) controller in addition to the popular PID controller is provided. The controllers’ efficiencies are compared in terms of server utilization, admission rate, drop rate of requests and the agility of the controllers. The results proved that the BELBIC offers faster rise time compared with the PID controller, which leads to better cloud utilization and a smaller number of dropped requests.