We document a substantial rise in the accuracy of U.S. firms’ expectations since the early 2000s, closely linked to firm-size dynamics and consistent with major advances in data-processing technologies. To study the macroeconomic implications, we develop a model of information production, in which information enables firms to optimize their scale, product choice, and pricing strategies. While information enhances the efficiency of resource allocation, it also facilitates price discrimination. The laissez-faire equilibrium is inefficient, warrants corrective policy interventions, and advances in data-processing technologies have ambiguous effects on social welfare. Calibrating our model to U.S. firm-level data, we find that data-processing advances have significantly increased TFP over the past two decades (5.3-6.7%), primarily by helping firms determine their optimal scale. Yet, the welfare benefits of these improvements have been modest (0.1-2.1%). Restricting data use, especially by large firms, could trigger larger welfare gains.