Abstract. Based on a dynamic general equilibrium model we study how the composition of technical progress, along three dimensions, affects transitional dynamics, with an emphasis on the speed of convergence. The three dimensions are, first, the degree to which technical change is embodied, second, the extent to which an endogenous source, learning, drives productivity advances, and, third, the extent to which the vehicle of learning is gross investment rather than net investment. The analysis shows that the speed of convergence, both ultimately and in a finite distance from the steady state, depends strongly and negatively on the importance of learning in the growth engine and on gross investment being the vehicle of learning rather than net investment. In contrast to a presumption implied by "old growth theory", a rising degree of embodiment in the wake of the computer revolution is not likely to raise the speed of convergence when learning by investing is the driving force of productivity increases.