Functionalization of semiconductor electrode surfaces with adsorbed 2-pyridinide (2-PyFI-*) has been postulated to enable selective CO2 photoelectroreduction to CH3OH. This hypothesis is supported by recent estimates of sufficient 2-Pyh(-)* lifetimes and low barriers for hydride transfer (HT) to CO2. However, the complete mechanism for reducing CO2 to CH3OH remained unidentified. Here, vetted quantum chemistry protocols for modeling GaP reveal a pathway involving HTs to specific CO2 reduction intermediates. Predicted barriers suggest that HT to HCOOH requires adsorbed HCOOH* reacting with 2-PyH-*, a new catalytic role for the surface. HT to HCOOH* produces CH2(OH)(2), but subsequent HT to CH2(OH)(2) forming CH3OH is hindered. However, CH2O, dehydrated CH2(OH)(2) easily reacts with 2-PyH-*, producing CH3OH. Further reduction of CH3OH to CH4 via HT from 2-PyH-* encounters a high barrier, consistent with experiment. Our finding that the GaP surface enables HT to HCOOH* explains why the primary CO, reduction product over CdTe photoelectrodes is HCOOH rather than methanol, as HCOOH does not adsorb on CdTe and. so the reaction terminates. The stability of 2-PyH-* (vs its protonation product DHP*), the relative dominance of CH2(OH)(2) over CH2O, and the required desorption of CH2(OH)(2)* are the most likely limiting factors, explaining the low yield of CH3OH observed experimentally.