Understanding the assembly of the most massive galaxies is one of the most important problems in astronomy today. It is now feasible to identify the most massive galaxies out to z 2.5, yet we have just begun to probe their evolution. To address this, we propose to obtain deep NIRI NIR long-slit spectroscopy of a sample of 15 ultra-massive galaxies (M_* > 10^11 M_0) at 1.8 < z < 2.6 with large MIPS 24 um and UV fluxes, for which we have unique, deep imaging with HST/NICMOS. With NIRI spectroscopy we will measure: H-alpha SFRs, combine SFRs with NICMOS-derived sizes to calculate the surface star formation density and to derive estimates of cold gas masses. This will be the first systematic study of these properties within ultra-massive galaxies at this redshift. With this new data, we will investigate how gas supplies, star formation activity, and AGN feedback mechanisms influence the early formation of massive galaxies at z 2.