A massive star appears to have exploded twice — first as a supernova, and then again days later when the two neutron stars its first explosion had created violently crashed into each other — in what astronomers announced in December 2025 may be the first event of its kind ever observed, so unprecedented that physicists have invented a new word for it: "superkilonova"

The unusual sequence began with a signal from a routine source. On 18 August 2025, the LIGO gravitational wave detectors in the United States and the Virgo detector in Italy registered ripples in spacetime characteristic of two compact objects merging — the kind of signal that has, since 2017, become the standard signature of binary neutron star mergers. The follow-up alert went out to the global astronomical community within minutes. Optical telescopes around the world, including the W. M. Keck Observatory in Hawaii, the Fraunhofer telescope in Germany, and the Zwicky Transient Facility at Palomar, slewed toward the position the gravitational wave detectors had triangulated. They found, at the predicted location, a bright new point of light in a galaxy approximately 1.3 billion light-years away. The object was catalogued as AT2025ulz. For the first three days of observations, it behaved exactly like the only previous confirmed kilonova in the history of astronomy — the 2017 event GW170817, which had been the first multi-messenger detection ever made and which had produced the red glow characteristic of newly synthesised heavy elements like gold and platinum.

According to Caltech’s official announcement of the discovery, the situation became substantially more interesting on day four. AT2025ulz did not fade out the way GW170817 had. Instead, the object brightened again. Its colour shifted from red toward blue. Its spectrum began to show hydrogen emission lines — a feature that kilonovae do not produce, but that supernovae routinely do. By the end of the week, several astronomers monitoring the source had concluded that AT2025ulz was simply an ordinary supernova that had happened to occur near the LIGO/Virgo localisation by coincidence, and that the gravitational wave signal was probably unrelated. Kasliwal and her collaborators were not among them. As Kasliwal described the moment in the Caltech announcement: “At first, for about three days, the eruption looked just like the first kilonova in 2017. Everybody was intensely trying to observe and analyze it, but then it started to look more like a supernova, and some astronomers lost interest. Not us.”

The interpretation

The Kasliwal team’s interpretation, developed in collaboration with the Columbia University astrophysicist Brian Metzger — whose theoretical work had been predicting events of this kind for more than a decade — proposes that AT2025ulz is neither a kilonova nor a supernova in isolation, but a hybrid of the two. As reported by Astronomy.com’s coverage of the proposed superkilonova mechanism, the proposed sequence runs as follows. A massive star, rotating rapidly, reaches the end of its nuclear fuel and undergoes core collapse, producing a supernova explosion. The collapse produces not one neutron star — the usual outcome — but two, either via fission of the collapsing core into two separate dense remnants, or via formation of a single neutron star surrounded by a disc of material that subsequently condenses into a second neutron star. The two neutron stars, formed essentially simultaneously and at very close range, immediately begin to spiral toward each other under their mutual gravitational attraction. Within hours, they merge — producing the gravitational wave signal LIGO and Virgo detected, and the kilonova that the optical telescopes observed during the first three days. The expanding shell of debris from the original supernova then catches up and overtakes the kilonova’s light, hiding the kilonova signature behind the slower-evolving supernova spectrum. By the time the supernova features appear in AT2025ulz, the kilonova has been buried inside the expanding supernova ejecta.

The key piece of evidence supporting this interpretation, beyond the unusual light-curve evolution, comes from the gravitational wave data itself. As covered by Space.com’s coverage of the LIGO-Virgo signal analysis, the gravitational wave signal indicated that at least one of the two merging objects was less massive than the Sun — substantially smaller than the typical neutron star, which weighs in between 1.4 and 2.5 solar masses. A neutron star less massive than the Sun is not impossible, but is rare, and is consistent with the unusual formation channel the superkilonova hypothesis requires. Additionally, supernovae from galaxies 1.3 billion light-years away are not expected to produce gravitational waves detectable by LIGO and Virgo at all; the supernova explosion process is generally too symmetric to generate strong gravitational radiation. The fact that LIGO and Virgo detected a clear signal coincident with AT2025ulz’s location strongly suggests that the gravitational wave source was a compact merger, not the supernova itself.

What remains uncertain

The Kasliwal team is explicit that the superkilonova interpretation is not yet confirmed. As reported by EarthSky’s coverage of the broader implications, the alternative explanation — that AT2025ulz is an ordinary supernova and the gravitational wave signal was either a chance coincidence or a different unrelated source — cannot be statistically ruled out from a single observation. The team’s analysis estimates that the probability of chance coincidence is low but not negligible. Confirming or refuting the superkilonova interpretation will require finding additional events of the same kind, in galaxies where the timing and spatial coincidence between gravitational wave signals, kilonova-like optical transients, and supernova-like later evolution can be cross-checked against the predictions of the Metzger theoretical model.

The instruments coming online over the next several years are well-positioned to provide that follow-up. The Vera C. Rubin Observatory in Chile, beginning routine operations in 2025-2026, will survey the entire visible sky every few nights with sensitivity substantially exceeding any previous wide-field survey, dramatically increasing the rate at which transient events like AT2025ulz are caught early in their evolution. NASA’s Nancy Grace Roman Space Telescope, scheduled for launch in 2027, will provide near-infrared sensitivity sufficient to characterise the spectra of distant kilonovae in detail. The next-generation gravitational wave detectors planned for the 2030s — Cosmic Explorer in the US and Einstein Telescope in Europe — will extend the range over which binary neutron star mergers can be detected, potentially capturing hundreds of comparable events per year. If even a small fraction of these turn out to follow the superkilonova pattern, the Kasliwal team’s interpretation will be confirmed. If none do, AT2025ulz will remain a peculiar one-off whose explanation remains contested.

What is no longer contested is that the universe contains explosions more complicated than the two existing categories — supernovae and kilonovae — were designed to describe. The 2017 GW170817 detection had established that kilonovae exist and that they are the source of the heavy elements in the universe. The 2025 AT2025ulz detection, if the superkilonova interpretation holds, will establish that some of those kilonovae are themselves triggered by, or hidden inside, the supernova explosions of the massive stars that produced the merging neutron stars in the first place. Kasliwal’s own characterisation of the implication is direct: “Future kilonovae events may not look like GW170817 and may be mistaken for supernovae.” The category boundary between the two kinds of explosion, in other words, may turn out to be substantially blurrier than astronomers had assumed. The next several years of observations will determine whether AT2025ulz is the first example of a new class of events or simply an unusually complicated single case.