The purpose of this paper is to evaluate the potentials of computers for building automationapplications. The criteria for this evaluation is to see if the computer can do more than theconventional control systems. This will be specifically reviewed for each of the many potential tasks of optimization. When it appears that the computer is superior for a particular task, thereasons for that will be listed, so that a quantitative cost-benefit analysis might be performedto check if the added investment is justified.