Astronomers have identified eight supermassive black holes so disproportionately massive relative to their host galaxies that none of the field’s seven leading computer simulations can account for their existence. The finding, led by the Max Planck Institute for Extraterrestrial Physics (MPE) and published in Astronomy & Astrophysics, dismantles one of cosmology’s most foundational assumptions: that black holes and their host galaxies grow together, in lock-step, across billions of years.

In a normally functioning galaxy, the central black hole accounts for roughly half a percent — about one part in 200 — of the galaxy’s total stellar mass, a remarkably consistent ratio observed across decades of local-universe measurements. The eight newly discovered systems shatter that picture: their black holes account for at least five percent of their host galaxies’ stellar mass — a ratio of 1:20 — more than ten times the typical value. The discovery was made using data from the eROSITA X-ray telescope, operating aboard the Spektr-RG space observatory, supported by optical spectroscopy from the Sloan Digital Sky Survey and archival ultraviolet-to-infrared photometry.

A Ratio That Has No Precedent in Any Major Model

The eight systems were identified in a 140-square-degree patch of sky — the eROSITA Final Equatorial-Depth Survey (eFEDS) field — containing 22,079 quasars. Quasars are galaxies whose central black holes are actively devouring surrounding material and releasing intense radiation across the electromagnetic spectrum; the X-ray emission captured by eROSITA confirmed that all eight of the identified systems are currently accreting matter. Three of the eight fell among the 200 brightest quasars in the survey’s deepest sub-field, a finding that statistical modeling confirms is not a coincidence of individual measurement errors.

“Astronomers assumed that central black holes always grow closely coupled to their host galaxies,” said Dr. Johannes Buchner, a postdoctoral researcher at MPE who led the study. “Our results show that this is not always the case.”

The black holes themselves are enormous by any measure — between roughly 800 million and four billion solar masses — which makes the largest of them approximately 1,000 times more massive than Sagittarius A*, the black hole at the center of our own Milky Way. Yet what makes these systems scientifically alarming is not their absolute size but their size relative to their host galaxies, which are strikingly faint.

“The faintness of the host galaxies shows that they cannot contain a stellar population comparable to that of a Milky-Way-sized galaxy,” Buchner said. “If they did, they would be much more clearly visible.”

Prof. Kirpal Nandra, Director of High-Energy Astrophysics at MPE and a co-author, offered an analogy that captures the absurdity of the finding: “Think of it like finding a Great Dane living in a studio apartment. The black hole has simply grown too large for its home.”

How Mass Is Measured Across Billions of Light-Years

Measuring both the black hole and the host galaxy with enough precision to establish the mass ratio required combining data from six telescopes spanning five wavelength ranges — and a spectroscopic technique that turns the width of a spectral emission line into a mass estimate.

The black hole masses were derived from optical spectra from SDSS, specifically using the broad H-beta emission line. That line originates from ionized gas orbiting the black hole in what astronomers call the Broad-Line Region — a structure too small to image directly at these distances but whose velocity can be read from the Doppler broadening of the emission peak. In these systems, the gas was traveling at roughly 40 million kilometers per hour (about 25 million miles per hour). Combined with estimates of the Broad-Line Region’s physical size — inferred from the black holes’ luminosity using well-established calibrations — this velocity yields a virial mass. Individual estimates carry an uncertainty of approximately a factor of three, but the study’s conclusions rest on the statistical properties of the population, not on any one measurement.

The host galaxy masses required a different approach. The team drew on ultraviolet observations from the GALEX satellite, optical imaging from the DESI Legacy Imaging Survey, near-infrared data from the VISTA Hemisphere Survey, and mid-infrared observations from the WISE satellite, modeling the quasar and galaxy light separately through a technique called galaxy-quasar decomposition. The result: faint host galaxies containing far fewer stars than a Milky Way-scale system would require to produce the observed brightness.

“The spectra reveal black holes with masses of roughly one to four billion Suns,” said Qiaoya Wu, a doctoral researcher at the University of Illinois Urbana-Champaign. “What is particularly unusual, however, is their mass relative to the stellar mass of their host galaxies.”

Still Growing — and Growing Faster Than the Galaxy Around Them

Perhaps the most striking element of the finding is that these black holes are not finished. The X-ray emission indicates an average black hole growth rate of roughly 40 million solar masses per billion years — a pace that could double their already extreme mass within approximately one billion years, making an already record ratio even more so.

“These black holes are already unusually massive compared with their host galaxies, yet they are still growing,” said Catarina Aydar, a doctoral researcher at MPE.

The central unanswered question is how a black hole could have accumulated so much mass without a proportional increase in its galaxy’s stellar mass. Current theoretical models propose two competing scenarios for how black holes and galaxies co-evolve: a causal one in which energetic outflows from the black hole physically regulate star formation in the surrounding galaxy, and a statistical one in which the observed correlation emerges through the averaging effect of repeated galaxy mergers over cosmic time. Both scenarios predict that the local-universe scaling ratio — 1:200 — is the equilibrium state that systems converge toward. The eight discovered systems are so far from that equilibrium, and so far along in cosmic history, that neither model has an obvious explanation for how they got there.

How Do You Get a Great Dane in a Studio Apartment?

Three growth scenarios are currently under active investigation, none yet confirmed:

Quenched star formation: The black hole’s energetic output may have shut down star formation early and permanently, leaving the galaxy stunted while the black hole continued to feed from remaining gas. The eROSITA data cannot directly confirm whether the host galaxies are still forming stars; higher-resolution imaging will be required.

Unusually massive seed: The black hole may have formed from a much heavier initial “seed” — possibly a direct-collapse black hole in the early universe — rather than from the collapse of a single massive star. This connects to findings from the James Webb Space Telescope (see below).

Super-Eddington accretion: Standard models assume black holes are limited in how fast they can grow by the Eddington luminosity — the point at which radiation pressure from infalling matter pushes back against further accretion. If these systems sustained prolonged periods of accretion above that limit, they could have grown far more rapidly than simulations allow.

The team compared their findings against seven leading cosmological simulations: Illustris, TNG, Horizon-AGN, EAGLE, SIMBA, Magneticum, and ASTRID. All seven predominantly produce black-hole-to-stellar-mass ratios near the local value of 1:200. None generates systems with mass ratios as extreme as those now observed. This simultaneous failure of all seven models is not easily explained by any single simulation’s known weaknesses — it points to a systematic gap in how the field’s best models handle a specific growth channel.

A Bridge to the Earliest Universe

The discovery connects directly to one of the most unsettling findings from the James Webb Space Telescope over the past several years: an abundance of apparently overmassive black holes at much higher redshifts, when the universe was less than a billion years old. Some of those systems show black-hole-to-stellar-mass ratios of 1:10 or even approaching 1:1 — far more extreme than the eight systems found here, but observed at cosmic epochs so early that rapid formation from unusual seeds seemed plausible as an explanation. A landmark 2024 paper in Nature by Maiolino et al. documented an overmassive black hole from when the universe was just 400 million years old, underscoring how pervasive the phenomenon appears to be.

The eROSITA result covers a very different cosmic epoch: redshifts of 0.3 to 0.8, corresponding to roughly three to seven billion years after the Big Bang — what cosmologists call “cosmic noon,” a period when the universe had already assembled much of its large-scale structure and the pace of galaxy evolution had slowed considerably. Finding overmassive black holes at this epoch suggests the phenomenon is not a relic of unusual physics at the very beginning of the universe. It may represent an ongoing growth channel that has operated across a substantial fraction of cosmic history.

“These results are exciting and unexpected,” said Prof. Roberto Maiolino, a professor at the University of Cambridge and a leading expert in black hole-galaxy co-evolution who was not involved in the study. “These observations indicate that black holes can become extremely massive relative to their galaxies, apparently to a large extent independently of the evolution of the stellar population. This challenges us to rethink how black holes grow over cosmic history.”

Why the Full Population May Be Much Larger

There is an important caveat to the scale of this finding: the eight discovered systems were found precisely because they are currently accreting — actively feeding, radiating in X-rays, and therefore detectable by eROSITA. A supermassive black hole that has completed its overmassive growth phase and returned to quiescence would not appear in this sample at all. The method selects for the active, observable fraction of what may be a substantially larger population of dormant overmassive black holes spread across these cosmic epochs.

The eight quasars identified in this study likely represent a tip of a much larger, as-yet-uncharacterized population. If dormant overmassive black holes are common at redshifts of 0.3 to 0.8, the simulation gap is almost certainly larger than the eight detected systems alone suggest.

What eROSITA Has Done — and What It Couldn’t Finish

The data for this discovery came from the eROSITA Final Equatorial-Depth Survey (eFEDS), a 140-square-degree calibration field observed during eROSITA’s performance verification phase in November 2019, before its formal all-sky survey program began. On July 31, 2026, the German eROSITA Consortium released eROSITA Data Release 2 (DR2), built from the first three complete all-sky scans and containing nearly two million X-ray sources — approximately double the number in the first public release — providing the most comprehensive catalog of the X-ray sky currently available.

However, eROSITA’s science operations were suspended on February 26, 2022, after Germany froze cooperation with Russia following the invasion of Ukraine, with the telescope having completed four of its planned eight all-sky survey passes. The remaining four passes — which would have increased the survey’s depth by roughly twofold and its sky coverage significantly — were never completed. Analysis of the existing German data continues, and future spectroscopic surveys including 4MOST and upcoming high-resolution imaging facilities will help determine how common overmassive systems are and what physical processes are driving their formation.

Frequently Asked Questions

How do scientists measure a black hole’s mass when they can’t see it directly?

The key is light from gas orbiting the black hole. When a black hole is actively feeding, the surrounding gas emits characteristic spectral emission lines — in this study, the H-beta line in visible light. The width of that line encodes the speed at which the gas is orbiting: broader line, faster orbit, more massive black hole. This “virial mass” technique has been calibrated against direct measurements in nearby galaxies and is the standard tool for measuring active black holes at cosmological distances. Individual estimates carry an uncertainty of roughly a factor of three, which is why the MPE team based its conclusions on the statistical properties of the full population rather than any single system.

Why does it matter that these eight black holes challenge simulations rather than just being unusual objects?

Because all seven simulations failed simultaneously. Each simulation uses different computational methods and different physical assumptions about how black holes grow and regulate their surrounding galaxies. If only one simulation had failed, the problem might lie in that specific model’s assumptions. When all seven — Illustris, TNG, Horizon-AGN, EAGLE, SIMBA, Magneticum, and ASTRID — fail to produce systems like these, the evidence points to a systematic gap in how the field’s best models handle a specific growth process. That is a much stronger signal than any single anomalous object would provide.

Could there be many more overmassive black holes that eROSITA can’t see?

Almost certainly. The detection method selects for black holes that are currently accreting actively and bright in X-rays. Any overmassive black hole that has completed its extreme growth phase and gone quiet — returned to the same dormant state as Sagittarius A* at the center of our galaxy — would be invisible in this survey. The eight discovered systems are by definition the active fraction of a potentially much larger total population. The finding is significant as it stands; the true scale of the gap between observation and current simulation models may be considerably larger once the quiescent population is accounted for.

What would it take to explain these systems theoretically?

Researchers are currently investigating three main possibilities: that the host galaxies had their star formation quenched very early (leaving the black hole to grow unimpeded while new stars stopped forming), that these black holes formed from unusually massive “seed” objects in the early universe, or that they sustained prolonged periods of accretion above the Eddington limit — faster than standard physics predicts as sustainable. None of these has been confirmed; all require updating the sub-grid physics prescriptions in cosmological simulations, and higher-resolution observations of the host galaxies will be needed to distinguish between them.