Reducing the heat generated in traditional electronics is a chief motivation for the development of spin-based electronics, called spintronics. Spin-based transistors that do not strictly rely on the raising or lowering of electrostatic barriers can overcome scaling limits in charge-based transistors. Spin transport in semiconductors might also lead to dissipation-less information transfer with pure spin currents. Despite these thermodynamic advantages, little experimental literature exists on the thermal aspects of spin transport in solids. A recent and surprising exception was the discovery of the spin-Seebeck effect, reported as a measurement of a redistribution of spins along the length of a sample of permalloy (NiFe) induced by a temperature gradient. This macroscopic spatial distribution of spins is, surprisingly, many orders of magnitude larger than the spin diffusion length, which has generated strong interest in the thermal aspects of spin transport. Here, the spin-Seebeck effect is observed in a ferromagnetic semiconductor, GaMnAs, which allows flexible design of the magnetization directions, a larger spin polarization, and measurements across the magnetic phase transition. This effect is observed even in the absence of longitudinal charge transport. The spatial distribution of spin currents is maintained across electrical breaks, highlighting the local nature of this thermally driven effect.