In this talk I will a describe a new scenario for early cosmology, where an inflationary deSitter phase is obtained with a ghost condensate. The transition to radiation dominance is triggered by the ghost itself, without any slow-roll potential. Density perturbations are generated by fluctuations around the ghost condensate and can be reliably computed in the effective field theory. The fluctuations are scale invariant as a consequence of the deSitter symmetries, however, the size of the perturbations are parametrically different than conventional slow-roll inflation, and the inflation happens at far lower energy scales. The model makes definite predictions that distinguish it from standard inflation, and can be sharply excluded or confirmed by experiments in the near future. The tilt in the scalar spectrum is predicted to vanish ns=1, and the gravity wave signal is negligible. The non-Gaussianities in the spectrum are predicted to be observable: the 3-point function is determined up to an overall O(1) constant, and its magnitude is much bigger than in conventional inflation, not far from the present WMAP bounds.