Abstract –Firm agglomeration is positively correlated with productivity, and it exhibits significant heterogeneity across industries. Yet, the connection between agglomeration and corporate investment remains underexplored. We develop a model of information sharing which predicts that knowledge-intensive industries and industries with more uncertainty benefit the most from agglomeration due to the subsequent improvement in project selection. Using counterfactuals that account for nonrandom location decisions and industry concentration, we find a strong positive relationship between (a) industry uncertainty/knowledge intensity and (b) the proximity of headquarters, patent inventors, and customer–suppliers. In addition, we exploit techniques de- signed to address inherent difficulties in estimating peer effects, and we find that investment externalities are positively related to proximity and uncertainty/knowledge intensity. This is consistent with the predictions of our model.