Key facts
- Black-hole growth is capped by the Eddington limit, where radiation from heated infalling gas balances gravity and stops faster infall; because the limit scales with mass, growth proceeds exponentially, says Zubovas.
- Assuming about 10 percent of infalling mass becomes radiation, a black hole's mass grows by a factor of e, roughly 2.7, every 45 million years, according to Zubovas.
- The second proposed escape is to break the Eddington limit by removing energy from infalling gas through channels other than radiation, enabling far faster super-Eddington growth, says Zubovas.
- A slowly spinning black hole radiates away only about 6 to 7.5 percent of swallowed mass instead of 10 percent, says Zubovas.
A black hole about 3 billion times the mass of the sun existed when the universe was less than 700 million years old, and astrophysicist Kastytis Zubovas says quasar J0313-1806 marks a two-decade puzzle over how supermassive black holes grew so large so quickly.
Nearly every massive galaxy hosts a central supermassive black hole, with local examples ranging from a few million to more than 10 billion solar masses and weighing about 1,000 times less than their host galaxies, according to Madisyn Brooks and colleagues. Stellar-mass black holes of a few to roughly 100 solar masses form when massive stars die in supernovae, yet how those seeds grew into supermassive black holes remains a mystery. Mergers, stellar consumption, and gas accretion can explain present-day supermassive black holes because they had more than 13 billion years to grow.
According to Brooks et al., JWST has found supermassive black holes in galaxies when the universe was less than a billion years old, and some are far more massive than expected for their hosts, reaching one-tenth of host-galaxy mass. The team used stacking—averaging spectra from surveys with no individual broad-emission-line detections, grouped by similar redshift and galaxy luminosity—to hunt faint accretion signals in distant galaxies. Broad emission lines appeared in five of 16 averaged spectra and more often at higher average redshift, indicating hidden active galactic nuclei in many high-redshift galaxies.
According to Brooks et al., the hidden black holes are on average less massive than supermassive black holes reported in other studies at the same redshift, with masses estimated from broad-line properties and host masses from spectral energy distribution fitting. They are generally under 1 percent of host-galaxy mass and possibly consistent with local supermassive black holes, implying a wider range of black-hole masses per galaxy mass at high redshift and a bias toward spotting the most massive. At lower redshift, individual AGN status can be determined definitively. The results fit either a light stellar-mass seed feeding at high but realistic rates or a heavier seed accreting more slowly, with the strongest constraints coming from the highest-redshift spectra and more data needed at even higher redshift.
Zubovas describes the Eddington limit as the point where radiation from heated infalling gas balances gravity and stops faster infall, with the limit scaling with mass so growth proceeds exponentially. Assuming about 10 percent of infalling mass becomes radiation, a black hole’s mass grows by a factor of e, roughly 2.7, every 45 million years. A hundred-solar-mass stellar black hole would need about 17 e-folding steps, roughly 770 million years, to reach 3 billion solar masses, and once seed-formation time is included less than 600 million years were available.
Zubovas says one proposed escape is a heavy seed of at least 10,000 solar masses formed by direct collapse of enormous gas clouds, skipping the slowest part of the growth climb. The second is to break the Eddington limit by removing energy from infalling gas through channels other than radiation, enabling far faster super-Eddington growth. Direct collapse demands an unusual set of conditions, and it remains unclear whether actual accretion flows can sustain super-Eddington rates across hundreds of millions of years.
Those caveats led Zubovas and Matas Tartėnas to investigate a third explanation that needs neither heavy seeds nor a broken speed limit. In a 2026 Astronomy & Astrophysics paper carrying DOI 10.1051/0004-6361/202558523, the pair tested chaotic accretion, an older idea that requires neither a heavy seed nor a broken limit. Both authors are theoretical astrophysicists at the Center for Physical Sciences and Technology in Vilnius, Lithuania.
Dr Kastytis Zubovas is a theoretical astrophysicist at the Center for Physical Sciences and Technology in Vilnius, Lithuania, studying how supermassive black holes affect host galaxies through large-scale outflows during active phases. Dr Matas Tartėnas is a theoretical astrophysicist at the same center who uses numerical hydrodynamical simulations to study supermassive black-hole feeding and feedback.
Zubovas says that in the chaotic-accretion model, a slowly rotating dark matter halo delivers gas from random directions, and each stream briefly feeds a small, randomly tilted disk. Competing spin-up and spin-down episodes leave the black hole’s spin low, since spin-down episodes are rarer but more efficient per unit mass. A slowly spinning black hole radiates away only about 6 to 7.5 percent of swallowed mass instead of 10 percent.
That lower efficiency shortens the growth timescale from 45 to roughly 27-32 million years, cutting the time needed to grow J0313-1806 from 770 million years to 550 million years, Zubovas says. Large-scale gas supply can exceed the Eddington rate, but the material reaching the event horizon stays near that limit while most of the rest is removed along the way, letting the black hole feed nearly continuously. Virial mass estimates out to redshift 10.6, including J0313-1806, fit a seed of about 100 solar masses or smaller formed around 100 million years after the Big Bang and growing on a roughly 32-million-year timescale.
Zubovas says the result shows chaotic accretion can work, not that alternatives fail, and several mass estimates remain unexplained because they rely on methods with more assumptions about accretion flow. Under this model, the heaviest early black holes should turn slowly, but prolonged super-Eddington feeding would spin them up quickly. ESA’s planned NewAthena observatory is best suited to measuring those spins but will not launch until at least 2037.
Zubovas says the model predicts maximum black-hole mass rises very rapidly until redshift 7.5 and then nearly plateaus, though current data hint at this plateau but cover too little volume to be conclusive. Euclid and Roman wide-area surveys should find enough black holes to test that prediction, and a robustly measured black hole over 1 billion solar masses at redshift 9.1 or higher would falsify the model. Once a rapidly growing black hole reaches 100,000 to 10 million solar masses, depending on its halo, accretion energy can drive outflows that clear surrounding gas and make it visible as a possible little red dot.
Zubovas concludes that lightweight stellar-mass seeds, feeding close to the Eddington limit at low spin, can account for the earliest supermassive black holes without requiring massive seeds or super-Eddington growth. Whether that actually happened is a question for next-generation telescopes, he says.
JWST and Gemini North found the oldest pair of supermassive black-hole-powered quasars ever seen, observed when the universe was 900 million years old—a thirteenth of its current age—with black holes as massive as present-day examples, said team leader Silvia Onorato of NOIRLab. The team says the universe was about 900 million years old when black holes of 1.5 billion and nearly 400 million solar masses existed 18,100 light-years apart—about one-fifth the diameter of the Milky Way—as quasar pair J1602+4228AB, nicknamed Nāpōwawāʻenakaulua, the most compact pair from Cosmic Dawn and the third such pair seen at that epoch.
Onorato’s team describes quasars as supermassive black holes actively feeding on surrounding matter, and radiation from a single quasar can outshine all the stars in its host galaxy. Onorato said there is no definitive theoretical explanation for even one giant quasar so early in the universe’s history, much less two.
Onorato said the brighter quasar had been known for more than two decades, and its fainter companion appeared in JWST’s Aether survey of 200 distant quasars. Gemini North on Maunakea resolved the two sources, less than one arcsecond apart with one far fainter, in two nights of clean, sharp weather. Spectra put the two objects at almost exactly the same cosmic distance, though gravitational lensing by an intervening galaxy cannot yet be completely ruled out; subtle spectral differences favor two distinct quasars, pending further analysis.
The team says both apparent quasars consume matter at around the Eddington limit, the maximum feeding rate models allow. The close pair’s frequency in the Aether survey implies a quasar concentration on small scales roughly 100 to 1,000 times higher than expected from extrapolated clustering estimates, though the figure depends heavily on one candidate and its assumptions.
Space reports that current models say black holes need at least 1 billion years to feed and merge into millions- or billions-of-solar-mass objects, yet JWST, operating since 2022, routinely finds quasar black holes as early as 400 million years after the Big Bang. The report says Nāpōwawāʻenakaulua could break current models.
The team says JWST’s Aether survey, using NIRSpec, has uncovered 200 quasars—just under half of all currently known early-universe quasars. The companion’s existence emerged while the researchers browsed the dataset, revealing that quasar J1602+4228A had a partner, J1602+4228B.
The results were published on Nov. 10 in The Astrophysical Journal Letters. Onorato called the system a unique laboratory for studying how the first massive galaxies and their central black holes assembled through direct galaxy interactions, and her team discovered the pair while examining JWST’s Aether survey data.





Join the discussion