I simulate the evolution of metabolism and mass to explain the curvature in the metabolic allometry for placental and marsupial mammals. I assume that the release of inter-specific competition by the extinction of dinosaurs 65 million years ago made it possible for each clade to diversity into a multitude of species across a wide range of niches. The natural selection of metabolism and mass was then fitted to explain the maximum observed body masses over time, as well as the current inter-specific allometry for metabolism. The estimated selection of mass specific metabolism was found to bend the metabolic allometry over time, with the strongest curvature in the placental clade. The rate of exponential increase in mass specific metabolism for placentals was estimated to 9.3x10^-9 (95% CI: 7.3x10^-9 - 1.1x10^-8) on the per generation time-scale. This is an order of magnitude larger than the estimate for marsupials, in agreement with an average metabolism that is 30% larger in placentals relative to marsupials of similar size.