Astronomers have traced the slowdown of thousands of supermassive black holes to a single dominant cause: the objects are consuming far less material than they did billions of years ago. A study drawing on X-ray data from roughly 1.3 million galaxies and about 8,000 actively growing black holes finds that feeding rates, not a shrinking population of active black holes or smaller black hole masses, explain why growth has declined sharply since a period known as “cosmic noon” around 10 billion years ago. The result reframes how scientists think about the long co-evolution of galaxies and the giant black holes at their centers.
What is verified so far
The central finding rests on a large, multi-telescope dataset. The research team combined X-ray observations from Chandra observations, the European Space Agency’s XMM-Newton, and the eROSITA instrument to build a census of black hole activity across cosmic time. That multi-wavelength approach allowed the researchers to separate the effects of black hole mass from the rate at which material falls onto each black hole, a distinction that single-band surveys struggle to make.
A key pillar of the analysis is the Chandra COSMOS-Legacy survey, a deep NASA program covering roughly 2.2 square degrees of sky. That extended exposure provided the X-ray detections, while optical and infrared counterpart identifications, documented in a supporting catalog, linked each X-ray source to its host galaxy. Those matched datasets gave the team the physical properties needed to test competing explanations for the growth decline.
The study systematically tested three possible drivers. Black hole growth could have slowed because fewer galaxies host active nuclei, because typical black hole masses shrank, or because the rate at which black holes consume gas dropped. The data pointed firmly to the third option. According to the study’s arXiv manuscript, the typical Eddington ratio, a measure of how fast a black hole feeds relative to its theoretical maximum, fell by roughly 1.35 dex between redshifts of about 1.5 to 2 and a redshift of about 0.2. In practical terms, that is approximately a 22-fold decrease in feeding efficiency across billions of years of cosmic history.
The authors implemented what they describe as a “wedding-cake” multi-field strategy, layering surveys of different depths and sky areas to capture both rare luminous objects and the more common faint ones. Wide, shallow observations provide large numbers of bright quasars, while narrow, ultra-deep pointings reveal the dimmer active nuclei that dominate the population by number. That design choice matters because previous work often relied on a single survey depth, which can bias results toward the brightest or the nearest active black holes while missing the broader population.
On the theoretical side, the team used standard accretion models to translate X-ray luminosities into mass growth rates. By combining those rates with estimates of black hole mass derived from host galaxy properties, they could track how efficiently black holes converted infalling gas into radiation at different cosmic epochs. The consistency of the inferred accretion histories across independent survey fields supports the robustness of the headline conclusion.
What remains uncertain
The publication status of the paper presents a minor but notable ambiguity. The preprint posted to arXiv’s help pages is described as accepted in The Astrophysical Journal, and a DOI (10.3847/1538-4357/ae173d) has been assigned. That identifier now resolves to the journal’s site, where the Astrophysical Journal article appears in the production system. Whether the fully typeset version is already live or still moving through final formatting is not entirely clear from the available records, but the existence of the DOI indicates that peer review has been completed.
A deeper open question is what physical mechanism caused the feeding rates to plummet. The study identifies the statistical pattern, showing that Eddington ratios dropped, but it does not claim to have pinpointed the astrophysical process responsible. Possible explanations circulating in the field include the depletion of cold gas reservoirs in maturing galaxies, energy feedback from earlier episodes of black hole activity that heated surrounding gas, and a decline in galaxy mergers that would otherwise funnel fresh material toward galactic centers. Each of these processes likely plays some role, but their relative importance remains to be quantified.
Disentangling these mechanisms will require follow-up observations at other wavelengths, especially those sensitive to cold gas and dust. Submillimeter measurements can map molecular gas in galaxies, while radio and infrared data can reveal jets and obscured star formation. Future facilities and surveys that complement the existing X-ray work will be crucial for turning the statistical trend into a detailed physical narrative of how galaxies and black holes shut down together.
The sample itself, while large, carries selection effects that the authors acknowledge through their survey design but that future work will need to refine. X-ray surveys preferentially detect unobscured or moderately obscured active black holes. Heavily buried objects, sometimes called Compton-thick sources, can escape detection even in deep exposures. If the fraction of hidden black holes changed over time, the measured decline in feeding rates could be partly an observational artifact rather than a purely physical trend. Correcting for this bias requires population synthesis models and, ideally, independent constraints from infrared and hard X-ray instruments that can pierce through thicker layers of gas.
There are also uncertainties associated with estimating black hole masses from host galaxy properties, especially at high redshift where galaxies are less settled and scaling relations are less secure. Any systematic shifts in those relations over time could subtly alter the inferred Eddington ratios. The study attempts to bracket these uncertainties by exploring different assumptions, but some residual model dependence is unavoidable.
How to read the evidence
Readers evaluating this result should distinguish between the primary observational evidence and the interpretive framework built around it. The raw X-ray detections and their optical counterparts are well-established data products curated through institutional archives and long-running survey collaborations. The statistical decomposition showing that Eddington ratios, rather than black hole counts or masses, drive the decline is a model-dependent conclusion, but one tested against multiple survey fields and telescope combinations, which strengthens it considerably.
Most press coverage of this result has emphasized the headline finding that black holes are “eating less,” without probing the limits of the claim. That framing is accurate as far as it goes, but it can obscure the fact that the study measures average population trends, not the behavior of individual objects. Some black holes at low redshift still accrete vigorously, and some at high redshift were already quiet. The power of the result lies in the shift of the entire distribution, not in a universal on-off switch.
One way to gauge the study’s reliability is to look at the consistency across independent datasets. The team did not rely solely on Chandra. By folding in XMM-Newton and eROSITA data, they could cross-check luminosity functions and accretion distributions across instruments with different sensitivities and sky coverage. Agreement across telescopes reduces the chance that any single calibration issue or field-specific anomaly is driving the conclusion, and it helps ensure that the observed decline in feeding rates is not an artifact of one instrument’s quirks.
A common assumption in popular accounts of black hole growth is that supermassive black holes spend most of their lives in a single, long-lived active phase. The new work instead supports a picture in which activity is episodic, with the duty cycle and intensity of accretion changing over cosmic time. At early epochs, gas-rich galaxies fuel frequent, near-maximal feeding episodes; at later times, the same black holes flicker on more sporadically and at lower intensities as their fuel supplies dwindle. Interpreting the study through this lens helps reconcile the strong average decline in accretion with the continued existence of rare, powerful quasars in the nearby universe.
The study also illustrates the value of open-access preprint servers in modern astrophysics. By the time the Astrophysical Journal article appeared with its peer-reviewed details, the core findings had already circulated widely via the arXiv posting, allowing other researchers to test related ideas, compare models, and plan follow-up observations. That rapid dissemination, supported by a network of member institutions and individual donors, has become a central part of how large, multi-year survey projects share their results with the broader community.
For non-specialists, the key takeaway is straightforward: the universe’s most massive black holes are not running out of room to grow, and they have not vanished; instead, the cosmic environment that once fed them so efficiently has changed. As galaxies aged and their gas supplies thinned, the average black hole’s diet shifted from a feast to a relative famine. Understanding that transition in detail will remain an active area of research, but the new analysis provides one of the clearest statistical roadmaps yet for how the universe’s brightest engines have dimmed over time.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
