In this paper we argue that the emergence of the dominant model of university organization, which is characterized by a large agglomeration of many (often loosely affiliated) small research groups, might have an economic explanation that relates to the features of the scientific production process. In particular, we argue that there are decreasing returns to scale on the level of the individual research groups, which prevent them from becoming to large, while we argue for positive agglomeration effects on the supra-research-group-level inside the university. As a consequence an efficient university organization would precisely consist of tying together many small individual research groups without merging them. Basing our empirical analysis on a multilevel dataset for German research institutes from four disciplines we are able to find strong support for the presence of these effects. This suggests that the emergence of the dominant model of university organization may also be the result of these particular features of the production process, where the least we can say is that this model is under the given circumstances highly efficient.