Recent calculations strongly suggest assuming an antiferromagnetic spin arrangement for the Si(100)-(2×1) reconstructed surface. The implications of this assumption for the alkali-covered surface are here studied by means of an extended Hubbard model. Whereas the substrate undergoes a gradual insulator-to-conductor transition with increasing coverage, the adlayer suddenly metallizes above a critical coverage value. Recent core-level data are not incompatible, as previously believed, with an ionic bond description of this interface.