In this note, we consider confining gauge theories in $D=2,3,4$ defined by $S^2$ or $T^2$ compactification of higher-dimensional conformal field theories with gravity duals. We investigate the behavior of these theories on de Sitter spacetime as a function of the Hubble parameter. We find that in each case, the de Sitter vacuum state of the field theory (defined by Euclidian continuation from a sphere) undergoes a deconfinement transition as the Hubble parameter is increased past a critical value. In each case, the corresponding critical de Sitter temperature is smaller than the corresponding Minkowski-space deconfinement temperature by a factor nearly equal to the dimension of the de Sitter spacetime. The behavior is qualitatively and quantitatively similar to that for confining theories defined by $S^1$ compactification of CFTs, studied recently in arXiv:1007.3996.

We present a new mechanism for slow-roll inflation based on higher dimensional supersymmetric gauge theory compactified to four dimensions with twisted (supersymmetry breaking) boundary conditions. These boundary conditions lead to a potential for directions in field space that would have been flat were supersymmetry preserved. For field values in these directions much larger than the supersymmetry-breaking scale, the flatness of the potential is nearly restored. Starting in this nearly flat region, inflation can occur as the theory relaxes towards the origin of field space. Near the origin, the potential becomes steep and the theory quickly descends to a confining gauge theory in which the inflaton does not exist as a particle. This confining gauge theory could be part of the Standard Model (QCD) or a natural dark matter sector; we comment on various scenarios for reheating. As a specific illustration of this mechanism, we discuss 4+1 dimensional maximally supersymmetric gauge theory on a circle with antiperiodic boundary conditions for fermions. When the theory is weakly coupled at the compactification scale, we calculate the inflaton potential directly in field theory by integrating out the heavy W-bosons and their superpartners. At strong coupling the model can be studied using a gravity dual, which realizes a new model of brane inflation on a non-supersymmetric throat geometry. Assuming there exists a UV completion that avoids the eta-problem, predictions from our model are consistent with present observations, and imply a small tensor-to-scalar ratio.