Late time failure of Friedmann equation

It is widely believed that the assumption of homogeneity is a good zeroth order approximation for the expansion of our Universe. We analyze the correction due to subhorizon inhomogeneous gravitational fields. While at early times this contribution (which may act as a negative pressure component) is perturbatively subdominant, we show that the perturbative series is likely to diverge at redshift of order 1, due to the growth of perturbations. So, the homogeneous Friedmann equation cannot be trusted at late times. We suggest that the puzzling observations of a present acceleration of the Universe, may just be due to the unjustified use of the Friedmann equation and not to the presence of a Dark Energy component. This would completely solve the coincidence problem.