Scientists have seen something spectacular unfolding in Andromeda, our neighboring spiral galaxy, located some 2.5 million light-years away from Earth. The spectacular part is actually what they didn’t see: instead of exploding as a bright supernova, a massive star there seems to have simply vanished.
This case of “now you see it, now you don’t” isn’t some cosmic magic trick; it appears to be a black hole being born, right before our far-gazing eyes. The observation may represent a failed supernova—a highly sought-after find because of a somewhat embarrassing fact: despite an ongoing renaissance in black hole studies that have revolutionized our understanding of these mysterious objects, even now, no one really knows how they form.
Catching a black hole just as it emerges from a dying star, scientists hope, might change that.
On supporting science journalism
If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Astronomers know that stars of about eight solar masses or more eventually collapse under their own weight when they run out of thermonuclear fuel in their core. The overlying layers fall inward, compressing the core into a city-sized ball of neutrons—a neutron star—and rebounding outward from the core in star-shaking shockwaves. The star explodes as a supernova when the shockwaves reach the surface, leaving the naked neutron star behind.
But for reasons that remain murky, within the most massive stars this obliterating shockwave sometimes fizzles, foiling the supernova. The star remains seemingly intact—until, doomed by gravity, its unstoppable implosion creates a black hole in its place.
Andromeda’s disappearing star wouldn’t be the first time astronomers have reported glimpsing a black hole born without a supernova, but it would be the closest, best candidate ever seen.
“I got goosebumps when I saw it disappearing into darkness.” — Kishalay De, astrophysicist
At least, that’s the conclusion of a study that was led by Columbia University astrophysicist Kishalay De and published today in Science. In 2022 De and his colleagues began searching for failed supernovae among nearby galaxies using archival data from NASA’s Near-Earth Object Wide-Field Infrared Survey Explorer (NEOWISE) mission, an infrared space telescope that mapped the sky from 2009 to 2024. In the fall of 2023 De found what he’d been looking for, nestled in Andromeda (also known as Messier 31, or M31): a yellow supergiant star about a dozen times heavier than our sun that had brightened starting in 2014 before it faded away, vanishing entirely from NEOWISE’s view by 2022. The team named the star M31-2014-DS1.
“I got goosebumps when I saw it disappearing into darkness,” De recalls.
Astronomers had estimated about once-per-century odds for a failed supernova to occur in any large spiral galaxy, so discovering one in the Milky Way’s next-door neighbor was almost too good to be true. Because Andromeda is such a popular astronomical target, De’s team could search the archives from a host of other world-class telescopes to confirm the star’s strange transition in infrared, optical and ultraviolet light. Stranger still, rather than seeing an outpouring of high-energy radiation from a newborn black hole feasting on stellar remains, Hubble Space Telescope data from 2022, as well as follow-up ground-based observations performed in 2023, revealed a dim, reddish blob where the star had once shone.
That blob, De says, was likely the dying star’s tenuously bound outermost layers wafting away as it collapsed. A minuscule portion of that material probably fell back, ablaze with x-rays as it trickled into the black hole; the rest would have formed an expanding, cooling shell of light-absorbing dust, faintly glowing from the otherwise-hidden fireworks that flared deeper within. “This is a prediction that’s been around for 50 years,” De says—and it’s supported by new observations with the James Webb Space Telescope (JWST) and the Chandra X-ray Observatory that he and his colleagues obtained in 2024. “When you look at the JWST data, it all just fits perfectly,” De says. (Those observations, however, arrived too late for inclusion in the researchers’ Science paper and have yet to be peer-reviewed.)
After intensive computational modeling of the star’s demise, De and his co-authors concluded that the failed supernova of M31-2014-DS1 produced a black hole of about five solar masses that was obscured by an ejected cloud of gas and dust with a tenth of the mass of our sun.
Their model also neatly explained puzzling observations of the previous best candidate for a failed supernova, NGC 6946-BH1. Discovered more than a decade ago by Ohio State University astrophysicist Christopher Kochanek and his colleagues, this object is harder to study because it’s about 10 times farther away—and thus about 100 times fainter—than M31-2014-DS1.
Kochanek—who was not part of De’s studies—agrees with the new results. Even so, he says, the team’s unifying model faces an inescapable problem in its reliance on dust—“one of the most notorious refuges of scoundrels in astronomy.” That is, dust is so versatile that it can be used as an explanatory catchall; for almost any set of confusing astronomical observations, some carefully contrived configuration of dust can be invoked to account for all its quirks. And this malleability cuts both ways.
Already, another peer-reviewed study has challenged De’s analysis—although neither he nor Kochanek find it convincing. Led by astrophysicist Emma Beasor of Liverpool John Moores University in England and published in mid-January in the Monthly Notices of the Royal Astronomical Society, it uses De’s JWST and Chandra data to reach a starkly different conclusion: M31-2014-DS1 and NGC 6946-BH1 alike could just as well be rare cases of dust-shrouded stellar mergers—two stars colliding and joining—rather than black-hole-birthing failed supernovae.
“Everyone’s excited to look for failed supernovae—me included,” Beasor says. “But I’d argue we don’t have models yet that make very clear predictions about what they should look like. So before we confirm one, what I want to do is rule out every other possible scenario that could cause a star to ‘disappear.’”
Time will tell which interpretation proves correct, Kochanek says, because one very clear model-independent prediction remains: a stellar merger will shine on, while a black hole will go dark. “And dust can’t obscure things forever,” he says. “In both of these scenarios, this dust is in an expanding shell, so the ‘veil’ will thin as the shell expands.”
That could take decades, however—potentially outlasting the planned lifetime of JWST, the best observatory, bar none, for such follow-up studies. That is why broadening the search to discover more candidates with next-generation facilities such as the ground-based Vera C. Rubin Observatory in Chile and NASA’s soon-to-launch Nancy Grace Roman Space Telescope could prove critical, says Suvi Gezari, an astronomer at the University of Maryland, who wasn’t part of these studies.
“Given those future capabilities, we can find more of these events to better determine if they’re all failed supernovae or stellar mergers or a mix,” she says. “Reading these papers, all I could think about was how exciting it is to be entering this era where we’ll get many, many more opportunities to study this phenomenon.”