The rotational kinematic Sunyaev-Zeldovich (rkSZ) signal, imprinted on the cosmic microwave background (CMB) by the gaseous halos (spinning "atmospheres") of foreground galaxies, would be a novel probe of galaxy formation. Although the signal is too weak to detect in individual galaxies, we analyze the feasibility of its statistical detection via stacking CMB data on many galaxies for which the spin orientation can be estimated spectroscopically. We use an "optimistic" model, in which fully ionized atmospheres contain the cosmic baryon fraction and spin at the halo's circular velocity vcirc, and a more realistic model, based on hydrodynamical simulations, with multi-phase atmospheres spinning at a fraction of vcirc. We incorporate realistic noise estimates into our analysis. Using low-redshift galaxy properties from the MaNGA spectroscopic survey (with median halo mass of 6.6 × 10 11 M ), and CMB data quality from Planck, we find that a 3σ detection would require a few×10 4 galaxies, even in the optimistic model. This is too high for current surveys, but upcoming higher-angular resolution CMB experiments will significantly reduce the requirements: stacking CMB data on galaxy spins in a ∼ 10 deg 2 can rule out the optimistic models, and ≈350 deg 2 will suffice for a 3σ detection with ACT. As a proof-of-concept, we stacked Planck data on the position of ≈ 2, 000 MaNGA galaxies, aligned with the galaxies' projected spin, and scaled to their halos' angular size. We rule out average temperature dipoles larger than ≈ 1.9 µK around field spiral galaxies.