LRDs could be "black hole stars" — objects where a growing black hole is embedded within and powered by a massive, bloated stellar envelope, essentially a quasi-star in the classical sense (Begelman, Volonteri & Rees 2006; Begelman, Rossi & Armitage 2008; Begelman 2010). Rather than being conventional AGN in compact galaxies, the entire emission region would be a single, exotic stellar-scale object. The most direct observational case for this interpretation is the recent discovery of MoM-BH*-1, a source whose spectrum is dominated by a dense gas-enshrouded black hole — what Naidu et al. (2025) explicitly dub a "black hole star" — and which reproduces the key features of a typical LRD when combined with a faint host galaxy contribution.

Why It's Attractive for LRDs

The observational puzzle of LRDs (Matthee et al. 2024; Greene et al. 2024) — very red rest-frame optical colors, broad Balmer emission lines, yet surprisingly blue UV slopes, and little to no X-ray or radio emission (Ananna et al. 2024) — is hard to reconcile with standard AGN+host galaxy models. The quasi-star picture offers a few natural explanations:

• The broad Balmer lines would arise from the dense, rapidly infalling envelope rather than a classical BLR, explaining their Balmer-heavy spectrum (sometimes showing Balmer absorption) without requiring a standard accretion disk.
• The lack of X-ray emission is easier to understand if the accretion occurs deep inside a stellar envelope that Compton-scatters and thermalizes hard photons before they escape — the effective photosphere radiates at much lower temperatures.
• The V-shaped SED (blue UV + red optical) could reflect the composite emission of the quasi-star's outer envelope dominating in the optical while the host galaxy's young stellar population contributes in the UV — a picture directly demonstrated in Naidu et al. (2025) and further supported by quasi-star spectral modeling in Campbell et al. (2025).

The Tension and Challenges

• Quasi-stars are theoretically expected to be very short-lived transient phases in early black hole growth, so catching so many of them at z ~ 4–8 would require either a very high formation rate or longer lifetimes than models predict.
• The inferred black hole masses from broad line widths are often 10⁷–10⁹ M☉, which is arguably too massive for the quasi-star phase, which is more naturally associated with seed BH growth in the 10³–10⁵ M☉ range — though Begelman & Dexter (2025) have argued LRDs may represent the late stage of the quasi-star process, where the BH has already accreted a substantial fraction of the total mass.
• Reproducing the full SED quantitatively with quasi-star models remains a work in progress — Campbell et al. (2025) show that a 10⁶ M☉ quasi-star broadly reproduces LRD continua, but no fully self-consistent atmosphere model yet fits all spectral features end-to-end.