A new preprint posted to arXiv (Moghni et al. 2026) provides some of the clearest observational evidence yet that 'dead' galaxies are not truly inert. Using integral-field spectroscopy from the MaNGA survey, the team analyzed 140 carefully selected red geyser galaxies—massive, quiescent systems with stellar masses around 10^10.5 solar masses that display large-scale, bi-symmetric patterns of ionized gas emission. These patterns are widely interpreted as low-level AGN-driven outflows.

The researchers traced cool neutral hydrogen gas (temperatures of 100–1000 K) via spatially resolved Na I D absorption lines, fitting double-Gaussian profiles to measure velocities and dispersions across the galaxies' disks. They report that roughly 70% of this cool gas is inflowing at a median speed of 47 km/s—only about 10% of the expected free-fall velocity—with relatively ordered kinematics (velocity dispersion roughly 40% of the stellar dispersion). Compared with a matched control sample of quiescent galaxies, red geysers show both higher detection rates of this cool gas (63% versus 40%) and reservoirs approximately 1.6 times larger in area.

Notably, the absorbing gas clouds appear young, with acceleration timescales around one million years and accretion timescales near 20 million years, suggesting they are transient features. The preprint is not yet peer-reviewed, relies on a sample size that becomes modest once subdivided (only 30% are radio-detected), and Na I D absorption can be affected by stellar contamination and does not uniquely locate gas in three dimensions—important limitations explicitly acknowledged by the authors.

This work builds directly on the original discovery of the red geyser population by Cheung et al. (2016, Nature), which first highlighted these galaxies as a common but previously overlooked manifestation of 'maintenance-mode' AGN feedback capable of suppressing star formation without violent quasar outbursts. Where the 2016 paper and subsequent coverage largely focused on the outflow side of the story, Moghni et al. shift attention to the inflow side and its environmental dependence. They find that radio-loud red geysers host inflowing reservoirs seven times larger than radio-quiet ones, while galaxies experiencing environmental interactions (mergers or flybys) host reservoirs 2.7 times larger than isolated systems.

The analysis connects these observations to cosmological hydrodynamical simulations such as IllustrisTNG (Nelson et al. 2019, MNRAS), which predict that AGN feedback must act in a self-regulated cycle: cooling gas rains inward, fuels weak accretion onto the central black hole, drives gentle outflows that reheat or expel material, and prevents wholesale star formation. Until now, simulations and observations have spoken past each other; the new MaNGA results supply the missing observational link by showing that the cool 'galactic rain' is more prevalent precisely in systems with ongoing low-level AGN activity.

What much existing coverage has missed is the regulatory elegance of this loop. Quenching is not a single event that permanently shuts galaxies down. Instead, even after star formation is largely suppressed, minor interactions with neighbors appear to top up the cool-gas supply, which in turn sustains just enough AGN activity to keep the galaxy red and dead over billions of years. This bridges a long-standing gap between theoretical models that require continuous fuel to explain AGN duty cycles and the observed quiescence of massive ellipticals.

The findings also carry implications for future multi-messenger surveys. If galactic rain is common, JWST and ELT observations of higher-redshift quiescent systems may reveal more frequent cool-gas inflows than currently expected, while radio facilities like SKA could map how these inflows correlate with faint AGN jets. The preprint therefore reframes quiescence as an active, cyclical process rather than a static graveyard state—illuminating one of the central regulatory mechanisms that shapes galaxy evolution across cosmic time.