For measuring the nature of dark energy, Type Ia supernovae (SN Ia) are currently limited by the poorly constrained redshift evolution of SN Ia intrinsic colors. Current constraints allow for a large, 10% bias in the measurement of the equation-of-state parameter for dark energy. Luckily, a SN's ejecta velocity is highly correlated with its intrinsic color, but uncorrelated with other properties such as luminosity. We propose to measure the ejecta velocity of a large sample of high-z SN Ia to measure their velocity distribution at high z, and constrain any potential evolution in intrinsic color. The SN will be discovered and photometrically followed by the Dark Energy Survey (DES), the only currently running high-z SN survey. With these data, we can improve our distance measurements of the spectroscopic subset by a factor of ~2 (making the effective statistical size of the sample 4 times larger) and reduce any potential bias related to the intrinsic color to a subdominant level for all SN Ia, even if they have no velocity data. This program is critical for the success of both the SN component and the entirety of DES. The results of this program will also be used to improve cosmological constraints using old data and will help plan future SN surveys such as LSST and WFIRST.