The surface energy of a solid measures the energy cost of increasing the surface area. All normal solids therefore have a positive surface energy(1)-if it had been negative, the solid would disintegrate. For this reason it is also generally believed that when certain ceramics can be found in a highly porous form, this is a metastable state, which will eventually sinter into the bulk solid at high temperatures. We present theoretical evidence suggesting that for theta-alumina, the surface energy is strongly dependent on the size of the crystallites, and that for some facets it is negative for thicknesses larger than similar to1 nm. This suggests a completely new picture of porous alumina in which the high-surface-area, nanocrystalline form is the thermodynamic ground state. The negative surface energy is found to be related to a particularly strongly adsorbed state of dissociated water on some alumina surfaces. We also present new experimental evidence based on infrared spectroscopy, in conjunction with X-ray diffraction and surface-area measurements, that theta-alumina has indeed very stable surface OH groups at high temperatures, and that this form of alumina does not sinter even at temperatures up to 1,300 K.