Neutral hydrogen surveys count the universe's fuel reserves, and the counts keep coming in high at the top. ALFALFA and successor surveys consistently detect more galaxies with very large HI masses than semi-analytic models and hydrodynamic simulations built on ΛCDM predict: the high-mass end of the HI mass function carries an excess of gas-rich giants the models cannot stock (Haynes et al. 2011; Maddox et al. 2015).
The deficit is feedback's collateral damage. To match the galaxy stellar mass function, the models must suppress star formation in massive halos with energetic AGN and supernova feedback, and the suppression is indiscriminate: the same heating and ejection that prevents overgrown stellar masses also overheats or expels the cold gas reservoirs, so the simulated universe holds few massive galaxies that are simultaneously gas-rich. The observed population requires a balance the models cannot strike, maintaining enormous cold reservoirs in massive systems without either converting them to stars or blowing them away, a three-way equilibrium between accretion, consumption, and ejection that tuned feedback prescriptions consistently miss on the gas-rich side.
The standing sharpens with the coming HI census: WALLABY and eventually SKA will measure the HI mass function's high end at orders of magnitude better statistics, fixing exactly how many impossible gas reservoirs the models must explain.