The role in prebiotic chemistry that Brønsted and Lewis sites, both present at the surface of common aluminosilicates, may have played in favoring the peptide bond formation has been addressed by ab initio methods within a cluster approach. B3LYP/6-31+G(d,p) free energy potential energy surfaces have been fully characterized for the model reaction glycine + NH3 --> 2-NH2 acetamide (mimicking the true 2 Gly --> GlyGly one) occurring on (i) a Lewis site, (ii) a Brønsted site, and (iii) a combined action of Lewis/Brønsted sites. Compared to the gas-phase (gp) activation free energy of 50 kcal/mol, the Lewis site alone reduces the gp barrier to 41 kcal/mol, whereas the activation by the Brønsted site dramatically reduces the barrier to about 18 kcal/mol. Nevertheless, formation of the prereactant complex in this latter case will rarely occur, since water will easily displace the glycine molecule interacting with the Brønsted site. However, if a realistic feldspar surface with neighboring Brønsted and Lewis sites is considered, the proper prereactant complex is highly stabilized by a simultaneous interaction with the Lewis and the Brønsted sites, in such a way that the Lewis site strongly attaches the glycine molecule to the surface whereas the Brønsted site efficiently catalyzes the condensation reaction, showing that the interplay between Lewis/Brønsted sites is an important issue. The free energy barrier computed for the realistic feldspar surface model is 26 kcal/mol. The role of dispersive interactions on the free energy barrier and the stabilization of the final product, not accounted for by the B3LYP functional, have been estimated and shown to be substantial. Speculations about further elongation of the formed dipeptide have been put forward on the basis of the relatively strong interaction energy of the formed GlyGly dipeptide with the aluminosilicate surface.