Every black hole that LIGO has detected came from a dead star. The masses range from a few solar masses to nearly a hundred, all above the Tolman-Oppenheimer-Volkoff limit — the minimum mass a neutron star can collapse into a black hole. Below roughly two solar masses, stellar physics can't produce a black hole. The Chandrasekhar limit says stars lighter than 1.4 solar masses end as white dwarfs. The core-collapse pathway requires a more massive progenitor. The lightest stellar black hole should weigh at least two to three times what the Sun weighs.
On November 12, 2025, LIGO Hanford, LIGO Livingston, and Virgo detected gravitational waves from a compact binary merger designated S251112cm. Both components appear to have masses below one solar mass. The chirp mass sits between 0.1 and 0.87 solar masses. The probability that at least one component is sub-solar exceeds 99 percent. No electromagnetic counterpart was found, and the probability of either component being a neutron star is less than 8 percent.
If this is real, the objects are too light to be stellar black holes.
Magaraggia and Cappelluti (arXiv 2602.21295, February 2026) explore the implication. If stellar physics can't make sub-solar black holes, something else did — and the leading candidate is the early universe itself. Primordial black holes form not from dying stars but from density fluctuations in the radiation-dominated epoch, when overdense regions of the primordial plasma collapse directly into black holes before any star exists. Phase transitions during the QCD epoch — when quarks condensed into protons and neutrons — naturally produce a broad mass spectrum with a sub-solar component that populates exactly the mass range where S251112cm sits.
The constraint is sharp: if S251112cm is astrophysical, then at least 4 percent of all dark matter in the universe must consist of primordial black holes. One event in 4.35 effective observing years, combined with the predicted merger rate from the PBH mass function, sets the floor. The true fraction could be much higher. And if some fraction of the heavier mergers already detected by LIGO also come from primordial rather than stellar origins, the gravitational-wave catalog contains objects from two completely different formation channels — one astrophysical, one cosmological — that we've been treating as a single population.
But the detection is marginal. The false alarm rate is roughly one per four years — compare this to GW150914's false alarm rate of less than one per 200,000 years. The observed merger rate of 0.23 events per year has error bars spanning nearly an order of magnitude. The PBH mass function depends on assumed formation parameters — lepton flavor asymmetry, QCD transition details — that are modeled, not measured. A single event at this significance is an interesting hint, not a confirmed discovery.
The paper is honest about this. The analysis shows what the detection implies if real, not that it is real. Future observing runs with improved sensitivity would either confirm a population of sub-solar mergers or constrain them out of existence. What makes the result worth attention is the asymmetry between the effort of verification and the magnitude of the implication. If S251112cm is noise, nothing changes. If it's real, then a fraction of the dark matter has been hiding in plain sight as sub-solar black holes formed before any star existed, and some of the gravitational-wave events already in the LIGO catalog may have been misclassified as stellar remnants when they were actually relics of the first microseconds of the universe.
The weight of the evidence is light. The weight of the implication is not.