This is a story about one star swallowing another, and also about an accounting formula. In the Andromeda Galaxy, a giant star about seven times the mass of the Sun spent nearly a decade slowly engulfing a much smaller companion. In October 2025 the companion reached the giant’s core and the whole thing blew. Astronomers have had a name for this process since the 1970s, “common envelope evolution,” and a formula for predicting how it ends. Until now nobody had seen it happen. As Jacek Krywko reports for Ars Technica, a team led by Columbia University astrophysicist Viraj Karambelkar followed the event from the decades before it began to the eruption at the end, and has published it in Nature.
The puzzle it addresses goes like this. Some pairs of stars orbit each other so closely that they can’t have formed there, so their orbits must have shrunk. The leading theory runs as follows. One star gets old and swells into a giant, and its outer layers engulf its partner. Now both stars sit inside one shared, loosely held envelope of gas. The smaller star drags through that gas, loses orbital energy to friction and spirals inward. If enough of the envelope gets blown off, you end up with a much tighter pair. If not, the two merge.
The catch is that it happens fast, which in astronomy means it barely happens at all. “It lasts for a few years, maybe a few decades,” Karambelkar said. “Stars evolve on time scales of millions of years. So, it’s really like a blip on the entire cosmic time scale.”

The team found its candidate by searching Andromeda for stars that were changing in brightness, using the WINTER infrared telescope at Palomar Observatory and data from NASA’s NEOWISE mission. One giant had been getting brighter for seven years. “When a star first swallows its companion, it has a very subtle effect on its brightness,” Karambelkar said. “It slowly brightens, and a lot of light comes out in the infrared.” In April 2025 the team published a paper proposing that this system, AT 2025abao, was an engulfment in progress. They couldn’t yet rule out a duller explanation, a single star with instabilities deep in its core, so they kept watching.
They also went back through the archives, pulling observations from about a dozen telescopes and surveys, the oldest being photographic sky plates from 1989. Until 2018 the star barely changed. Models of those quiet years showed a giant late in life, about 350 times wider than the Sun. The companion couldn’t be seen at all: “The giant star outshines the smaller star by a factor of about 18,000,” Karambelkar said. So the team estimated its mass indirectly, from how much dust the system had shed before the engulfment, and came up with roughly 0.7 solar masses. That’s a ten-to-one mismatch with the giant.
Between 2018 and early 2025 the giant got two to three times brighter in visible light. It swelled, but its temperature stayed nearly constant. Karambelkar compares it to a spacecraft reentering the atmosphere. The companion skims through the giant’s outer layers, its orbit shrinks, and friction heats the envelope. What makes this convincing is that the books balance: the energy the drag should release matched the extra light the team measured.
Then the visible light fell by half while the infrared kept climbing. NASA’s SPHEREx telescope happened to catch the system in July 2025 and found the giant had grown to about 860 times the Sun’s width, roughly 2.5 times its quiet size, and had wrapped itself in fresh dust.

In October 2025 it got roughly 100 times brighter in 25 days and peaked at about 1.5 million Suns. “At one point, the rate at which the orbit was decaying became very, very fast,” Karambelkar said. “The companion just fell very quickly, very deep inside the envelope of the star, and it crashed into the core. That violent crash set off a bomb inside the star, and that explosion caused it to brighten very quickly and eject some of the material.”
The outburst looked like a luminous red nova, a kind of eruption already known to come from merging stars, though the researchers hadn’t expected one. “Honestly, our minds were blown,” Karambelkar said. The unusual part was the giant. Every previously known luminous red nova came from a younger star with dense, tightly bound outer layers, and that setup always ends in a merger. This was an old giant with a loose, churning envelope, which is exactly the kind of system that can leave a surviving tight binary behind. So did the companion survive?
The energy budget
To predict how these events end, astronomers use a formula with an efficiency parameter. It’s basically an allocation rule: of all the orbital energy the sinking companion gives up, what share goes into blowing the envelope away? AT 2025abao shows the answer changes as things go on. During the slow spiral, most of the energy left as light. In the final plunge the companion released about a hundred times more energy, but only around 1 percent of that was radiated. Nearly all the rest went into throwing a tenth of the envelope outward at about 10 times the speed needed to escape the giant’s gravity. The other nine-tenths stayed put, which suggests the companion merged with the core and didn’t survive as a tight partner.

“Initially, when the rate of energy deposition is low, the star is able to puff up and adjust to this new energy source inside it,” Karambelkar said. “But once the energy deposition rate increases very quickly as the companion spirals even closer, the star can no longer adjust itself. At that point, it needs to throw out some of its material.”
Here’s the fun part. The efficiency the team calculated matches values inferred from known post-engulfment binaries, so the formula passes. But the formula assumes the energy goes mostly into expelling a lot of envelope, and here it went into flinging a little envelope very fast. The total comes out right and the breakdown is wrong. That makes the formula a decent rule of thumb that misses some of the physics. Anyone who has built a financial model that hit the right answer for the wrong reasons will recognise the feeling.
There’s a side result too. Gas shed during the slow spiral piled up in the plane of the orbit, so the blast was blocked around the equator and could expand freely toward the poles. Karambelkar thinks that explains the hourglass-shaped nebulae seen around dying stars: “It had been an idea for a very long time that those hourglass shapes are shaped by binary interactions. It seems like what we’re seeing here confirms that hypothesis.”

Pending audit
The team says it can’t yet be sure the stars merged, because dust made from the ejected material hides whatever is left. “We want to continue monitoring the system for several years, possibly even decades, to watch the dust clear out and whatever is remaining inside emerge,” Karambelkar said. Not everyone is waiting to take the other side. A separate preprint, “Unravelling the Nature of AT 2025abao”, argues that the infrared precursor came from a companion of about 1.4 solar masses, roughly double the Columbia estimate, spiralling through the wind of a giant of about 6.9 solar masses. It suggests the event may be the first common envelope episode caught in the act of forming a close compact binary. In other words, survival. The James Webb Space Telescope, which took part in the observations, looked at the system just 47 days after the eruption peaked, which Karambelkar called the earliest follow-up ever of an event like this.
For now, Andromeda is as far away as today’s telescopes can spot these events. The Vera Rubin Observatory and the Nancy Grace Roman Space Telescope should be able to find them in about 100 Milky Way-like galaxies within some 50 million light-years. “We can have an influx of several of these events per year,” Karambelkar said.
So after half a century, the formula has finally been checked against an event as it happened, and not just against the binaries that came out the other side. The result: the top-line number holds, the line items don’t, and the one thing everyone wants to know, whether anybody survived, will have to wait until the dust clears.
