It has been recently suggested that the Bose-Einstein condensate formed by excitons in the dilute limit must be dark, i.e., not coupled to photons. Here, we show that, under a density increase, the dark exciton condensate must acquire a bright component due to carrier exchange in which dark excitons turn bright. This however requires a density larger than a threshold which seems to fall in the forbidden region of the phase separation between a dilute exciton gas and a dense electron-hole plasma. The BCS-like condensation which is likely to take place on the dense side, must then have a dark and a bright component -which makes it "gray". It should be possible to induce an internal Josephson effect between these two coherent components, with oscillations of the photoluminescence as a strong proof of the existence for this "gray" BCS-like exciton condensate. The carrier mass being very light and Coulomb attraction quite reduced by the large crystal dielectric constant, the exciton Bohr radius a X is two orders of magnitude larger than the hydrogen atom Bohr radius. So, manybody effects controlled by the dimensionless parameterD where N is the exciton number, L the sample size and D the space dimension, can easily be made significant for N not too large. The most dramatic one is the Mott dissociation of excitons into an electronhole plasma [2] when the distance between two excitons is of the order of their size. Actually, η ∼ 1 most often fall in an instability region with phase separation between a dilute exciton phase and a dense electron-hole phase. In Si and Ge [3,4], this exciton dissociation is spontaneous at T = 0 because the lowest energy phase is the high density electron-hole plasma which is stabilized by the multivalley structure of the conduction band. In the case of direct gap semiconductors, a similar phase separation can occur at T = 0, but only under a density increase [5,6].The exciton composite nature also appears through the fact that excitons come in "bright" and "dark" states. Bright excitons, coupled to σ ± photons, are made of (∓1/2) conduction electrons and (±3/2) valence holes. Carrier exchanges however transform two opposite spin bright excitons (+1) and (-1) into two dark excitons (+2) and (-2), these (±2) excitons being made of (±1/2) conduction electrons and (±3/2) valence holes. These dark excitons have actually a lower energy than bright excitons. Indeed, in addition to the intraband Coulomb processes responsible for the dominant part of the exciton binding energy, interband Coulomb processes also exist in which one conduction electron returns to the valence band while one valence electron jumps in a conduction state. By just writing [7] that the electron conserves its spin when it changes from a conduction state with orbital momentum l = 0 to a valence state with orbital momentum l = 1, it is possible to show that the electron-hole pair which undergoes these interband processes must be bright. Since Coulomb interaction between electrons is repulsive, bright excitons thus have an energ...