Using three-dimensional, moving-mesh simulations, we investigate the future evolution of the recently discovered gas cloud G2 traveling through the galactic center. From our simulations we expect an average feeding rate onto Sgr A* in the range of (5−19) × 10−8M⊙ yr−1 beginning in 2014. This accretion varies by less than a factor of three on timescales ∼ 1 month, and shows no more than a factor of 10 difference between the maximum and minimum observed rates within any given model. These rates are comparable to the current estimated accretion rate in the immediate vicinity of Sgr A*, although they represent only a small (< 10%) increase over the current expected feeding rate at the effective inner boundary of our simulations (racc = 750 RS ∼ 1015 cm). We also explore multiple possible equations of state to describe the gas. In examining the Br-γ light curves produced from our simulations, we find that all of our isothermal models predict significant (factor of 10) enhancements in the luminosity of G2 as it approaches pericenter, in conflict with observations. Models that instead allow the cloud to heat as it is compressed do better at matching observations.