The fossil record of marine invertebrates has long fueled the debate on whether or not there are limits to global diversity in the sea1–4. Ecological theory states that as diversity grows and ecological niches are filled, the strengthening of biological interactions imposes limits on diversity5–7. However, the extent to which biological interactions have constrained the growth of diversity over evolutionary time remains an open question1–4,8–12, largely because of the incompleteness and spatial heterogeneity of the fossil record13–15. Here we present a regional diversification model that reproduces surprisingly well the Phanerozoic trends in the global diversity of marine invertebrates after imposing mass extinctions. We find that the dynamics of global diversity is best described by a diversification model that operates broadly within the exponential growth regime of a logistic function. A spatially resolved analysis of the diversity-to-carrying capacity ratio reveals that only < 2% of the global flooded continental area exhibits diversity levels approaching ecological saturation. We attribute the overall increase in global diversity during the Late Mesozoic and Cenozoic to the development of diversity hotspots under prolonged conditions of Earth system stability and maximum continental fragmentation. We call this the "diversity hotspots hypothesis", which is proposed as a non-mutually exclusive alternative to the hypothesis that the Mesozoic marine revolution led this macroevolutionary trend16,17.