The discovery of several super-massive black holes (SMBHs) with masses of ~10^9 Msun at redshift as high as z=7.5 in wide area near-IR surveys challenges current models of black hole formation and early growth. However, these monsters represent the tip of the underlying SMBH population, both in terms of mass and luminosity. Deep X-ray surveys provide unprecedented access to the bulk of the population of accreting SMBHs at high redshift. I will present our recent results on the 33 AGN. In particular, we derived a large fraction (50-80%) of heavily-obscured (logNH>23) AGN, which does not evolve significantly from z=3 to 6. In contrast to low-redshift findings, the obscured AGN fraction does not appear to decrease significantly at high luminosities. We also found that the mass growth of SMBH at high- redshift is dominated by the fast and short AGN phase, with a possible low-rate, continuous accretion in galaxies playing a secondary role. Finally, I will show our findings on the high-redshift AGN X-ray luminosity function, focussing in particular on the slope of the faint end, accessible only by the deepest X-ray surveys. This is particularly important to assess the contribution of AGN to the cosmic reionization. All of these results will be placed in the context of SMBH seeds formation and growth, and comparison with expectations from simulations will be provided. I will also discuss how future X-ray missions like Lynx and Athena, together with new optical facilities like JWST and WFIRST, will boost our knowledge of the SMBH formation and evolution in the early universe.