The neoclassical growth model assumes fixed labor supply and competitive labor markets. Is it harmless to ignore monopsonistic power in the neoclassical growth model? The paper argues that it is not, especially if a growth model needs to be consistent with the long-run dynamics of the labor share. This paper solves a minimalist growth model with monopsonistic power at the firm level and two production technologies with different degrees of efficiency. The paper shows that monopsonistic power by the representative firm implies either a "level" or a "growth" effect in the determination of the labor share. If the two sectors feature unbalanced growth, the economy converges to a an asymptotic balanced growth in which the labor share asymptotically decline, in line with secular evidence on labor share dynamics. The paper shows also that the monopsonistic equilibrium has sizeable "misallocative" effects, since it implies the use of less efficient technologies that are not used by the optimal growth problem. Finally, the paper shows that the negative welfare effect of monopsony is larger when the model accounts for endogenous labor supply as the redistribution from wages to profits induces a reduction in hours worked. The generalized model is also consistent with recent evidence on balanced growth with declining labor supply.