In the last few decades, mainstream industrial organization has developed very powerful tools for analyzing firm behavior within a fairly narrow framework. In this essay, I argue that while these tools are important, their development has crowded out other important aspects of firm behavior from the mainstream of IO. I conclude that in order to continue to grow the power of IO to predict firm behavior, the field will need to embrace aspects of firms that have moved away from IO in recent years, including innovation, boundaries of the firm, regulation, managerial incentives and decision biases, and others.