This paper presents a model of growth through technical progress. The nature and scope of what is learned is derived from a set of axioms, and optimal search behavior by agents is then analyzed. Agents can search intensively or extensively. Intensive search explores a technology in greater depth, while extensive search yields new technologies. Agents alternate between these two modes of search. The economy grows forever and the growth rate is bounded away from zero. The growth rate is on average higher during periods of intensive search than during periods of extensive search. Epochs of higher growth are initiated by discoveries that call for further intensive exploration. This mechanism is reminiscent of the process described by Schumpeter as causing long-wave business cycles. Serial correlation properties of output and growth stem from the presence of intensive rather than extensive search. The two key parameters are technological opportunity and the cost of the extensive search.