The answer to this question is highly definition-dependent. If you do not work with the old hybridization model, but instead the full MO set, you will get three σ orbitals in the valence, of which one is antibonding, not just one:
Article The Rotational Barrier in Ethane: A Molecular Orbital Study
The questions of definitions involve: do you only consider the σx orbital or also the σs and σs* into your consideration? Do you set the cutoff for the "end" of the orbital at an expected value of half the electron density, a quarter or less?