Hard and soft spectral states of ULXs

I discuss some differences between the observed spectral states of ultraluminous X-ray sources (ULXs) and the canonical scheme of spectral states defined in Galactic black holes. The standard interpretation of ULXs with a curved spectrum, or a moderately steep power-law with soft excess and high-energy downturn, is that they are an extension of the very high state, up to luminosities ~ 1 to 3 L_{Edd}. Two competing models are Comptonization in a warm corona, and slim disk; I suggest bulk motion Comptonization in the radiatively-driven outflow as another possibility. The interpretation of ULXs with a hard power-law spectrum is more problematic. Some of them remain in that state over a large range of luminosities; others switch directly to a curved state without going through a canonical high/soft state. I suggest that those ULXs are in a high/hard state not seen in Galactic black holes; that state may overlap with the low/hard state at lower accretion rates, and extend all the way to Eddington accretion rates. If some black holes can reach Eddington accretion rates without switching to a standard-disk-dominated state, it is also possible that they never quench their steady jets.