A small red star is pulling gas from a brown dwarf as the pair circles every 87 minutes, packed into an orbit that could fit inside the Sun. Material strikes the star’s surface and lights up a hot spot, marking a meal that could continue for billions of years.

The system, ZTF J0440+2325, lies a few hundred light-years away. Aaron Householder and colleagues at MIT and other institutions described it in a study published in Nature Astronomy on October 5, 2026 (earlier version free on arXiv). Theory had long allowed a brown dwarf to feed a normal, hydrogen-burning star steadily, without being swallowed, but no one had seen it happen until now.

A strange triangle

The first clue was a shape on a graph. In observations from the Zwicky Transient Facility, which tracks changes in the sky from California’s Palomar Observatory, the system’s brightness kept tracing a triangle.

It brightened, faded and repeated the pattern. Kevin Burdge, now an assistant professor of physics at MIT, noticed it while searching through the survey’s data several years ago.

“I remember first looking at this and thinking: Stars don’t make triangular waveforms like this,” Burdge told MIT News.

The signal was clear enough to attract attention, but its cause remained unexplained for years. Eventually, Burdge and Householder returned to it, using observations from several telescopes to work out what could brighten and dim so dramatically every 86.65 minutes.

Solving the mystery

One early possibility was a black widow system. In these binaries, a rapidly spinning neutron star, the dense remnant of a dead star, bombards a much smaller companion with radiation.

That can make the companion’s heated side brighten as it turns toward us, producing a repeating signal. But a neutron star also has a strong gravitational pull, and the motion measured here did not fit that picture comfortably.

Astronomers tracked shifts in the system’s spectrum to measure the visible star’s movement toward and away from Earth. Its orbital velocity had a semi-amplitude of about 25 kilometers per second, much smaller than expected if a neutron star were tugging it around.

A neutron star could reproduce such a small signal only if the orbit were viewed almost face-on, an unlikely arrangement. Observations also found no X-rays at the levels typically associated with black widow systems.

Together, those clues strongly disfavored the neutron-star explanation. The spectrum instead revealed an M8 red dwarf, a small, cool star, with a low-mass companion consistent with a brown dwarf.

The team’s estimates put the red dwarf at roughly 85 Jupiter masses and its companion at around 25. A brown dwarf lacks enough mass to sustain the hydrogen fusion that powers a main-sequence star, placing these two objects on opposite sides of that boundary.

How the eating works

The brown dwarf sits so close to the red dwarf that it fills its Roche lobe, the region around it within which gas can remain gravitationally bound to it. Material can then spill through a point between the two objects and flow toward the star.

Astronomers call this transfer accretion. Here, the stream hits the red dwarf directly rather than settling into a disk around it.

The impact heats a patch of the star’s atmosphere. As the pair orbits, that bright region rotates into and out of view, producing the repeated rise and fall that first caught Burdge’s attention.

Computer simulations supported this explanation: particles leaving the brown dwarf followed paths straight onto the star. The changing visibility of the impact region broadly matched the observed brightness pattern, though the authors note that their hot-spot models do not reproduce every detail of it.

MIT estimates the transfer rate at about one hundred-thousandth of Earth’s mass each year, nearly two trillion kilograms every second.

That sounds ravenous, but the brown dwarf holds nearly 8,000 Earth masses. Even this enormous stream removes only around one part in a billion of its material each year.

How long will it last?

Billions of years is a possibility, not a measured countdown. Astronomers can observe the transfer happening now, but estimating its future means modeling how the objects and their orbit respond as material changes hands.

The researchers’ preprint gives a likely minimum timescale of roughly 10 million to 100 million years, with hundreds of millions to several billion years also plausible. Those figures describe a broad range of possible evolution.

A related theoretical study from the same research group, led by Samuel Whitebook, predicts lifetimes of around 100 million years for mass-transferring ultracool binaries under its model assumptions. That is useful context, but it is not an individual expiry date for ZTF J0440+2325.

The future depends on how efficiently the system loses orbital angular momentum and how the brown dwarf responds to losing mass. A simple calculation that divides its present mass by today’s feeding rate cannot capture all those changes.

What makes the discovery unusual is the evidence for sustained transfer. The brown dwarf has reached an arrangement in which losing material need not immediately lead to its destruction.

A second possible case

The published paper also describes ZTF J1444+4820, whose close pair completes an orbit every 67.16 minutes. Its brightness varies strongly with wavelength, and sharp eclipses suggest that a bright emitting region is periodically being hidden.

But the authors classify it as a strong candidate for a similar low-mass, mass-transferring system. Its components have not been established as securely as those in ZTF J0440+2325.

A brighter M4 star dominates its spectrum, obscuring the light needed to identify the close pair. The researchers interpret that star as a likely third member in a wider orbit.

The eclipses help test explanations for the changing brightness, but they do not settle the identities of the objects responsible. More detailed observations are needed before this pair can carry the same claim as the 87-minute system.

More to come

The researchers expect the Vera C. Rubin Observatory’s Legacy Survey of Space and Time to help uncover more systems like these. Its observations across six wavelength bands should be particularly useful for finding objects whose brightness changes much more strongly in some colors than others.

A larger sample could help explain how such small objects reach these extremely tight orbits. It would also give astronomers more examples against which to test their estimates of stability and lifetime.

The same group has already found a close cousin. In March 2026, a team led by Whitebook, with Burdge and Householder among the authors, reported ZTF J1239+8347 in The Astrophysical Journal Letters: one brown dwarf transferring material onto another in a 57.41-minute orbit. That pair contains two brown dwarfs, making it a different arrangement from the red dwarf and brown dwarf described here.

A slower ending

Astronomers predict that Earth will be engulfed when the Sun expands into a red giant billions of years from now. This distant pair shows that consumption can take a slower course, with a companion remaining in orbit while gradually losing itself. Every 87 minutes, the bright patch turns back into view, lit by another stream of gas arriving from the brown dwarf.