We constrain the ratio of black hole (BH) mass to total stellar mass of type-1 AGN in the COSMOS survey at 1<z<2. For 10 AGN at mean redshift z~1.4 with both HST/ACS and HST/NICMOS imaging data we are able to compute total stellar mass M_(*,total), based on restframe UV-to-optical host galaxy colors which constrain mass-to-light ratios. All objects have virial BH mass-estimates available from the COSMOS Magellan/IMACS and zCOSMOS surveys. We find zero difference between the M_BH--M_(*,total)-relation at z~1.4 and the M_BH--M_(*,bulge)-relation in the local Universe. Our interpretation is: (a) If our objects were purely bulge-dominated, the M_BH--M_(*,bulge)-relation has not evolved since z~1.4. However, (b) since we have evidence for substantial disk components, the bulges of massive galaxies (logM_(*,total)=11.1+-0.25 or logM_BH~8.3+-0.2) must have grown over the last 9 Gyrs predominantly by redistribution of disk- into bulge-mass. Since all necessary stellar mass exists in the galaxy at z=1.4, no star-formation or addition of external stellar material is required, only a redistribution e.g. induced by minor and major merging or through disk instabilities. Merging, in addition to redistributing mass in the galaxy, will add both BH and stellar/bulge mass, but does not change the overall final M_BH/M_(*,bulge) ratio. Since the overall cosmic stellar and BH mass buildup trace each other tightly over time, our scenario of bulge-formation in massive galaxies is independent of any strong BH-feedback and means that the mechanism coupling BH and bulge mass until the present is very indirect.