Silica surface functionalization is often done through the condensation of functional silanes on silanols, silica surfaces' terminal groups. APTES, aminopropyltriethoxysilane, is widely used due to its assumed high reactivity with silanols, kinetically promoted by the catalytic action of the terminal amine function. Here, we revisit, based on a quantitative analysis by solid-state 29Si NMR, the assembly of this silane on silica surfaces to investigate whether its presence results from grafting, i.e., hetero-condensation with silanol groups or from homo-condensation of silane molecules in solution leading to polycondensates physisorbed on silica. We investigate the interaction of APTES with a crystalline layered silicate, ilerite, and with amorphous nonporous silica. We also studied a second silane, cyanopropyltrichlorosilane (CPTCS), terminated with a nitrile group. Our results undoubtedly prove that while CPTCS is grafted on the silica surface, the presence of APTES on silica and silicate materials is only marginally associated with silanol consumption. The analysis of the signal related to silicon atoms from silanes (Tn species) and those from silica (Qn species) allowed for the accurate estimation of the extent of homo-condensation vs grafting based on the ratio of T-O-T/Q-O-T siloxane bridges. These findings deeply question the well-established certainties on APTES assembly on silica that should no longer be seen as grafting of alkoxysilane by hetero-condensation with silanol groups but more accurately as a homo-condensed network of silanes, predominantly physisorbed on the surface but including some sparse anchoring points to the surface involving less than 6% of the overall silanol groups.