JWST lensed-quasar survey hints at self-interacting dark matter on small scales

A new arXiv preprint analyzing 29 quasars whose light has been split into four images by foreground galaxies reports a population-level preference for self-interacting dark matter over standard cold dark matter at small cosmic scales. If confirmed, the result would be a major development in cosmology. But the paper has not yet been peer-reviewed, and its statistical case amounts to a hint rather than a definitive detection.

The paper, titled “JWST lensed quasar dark matter survey V: Hints of self-interacting dark matter from 29 quadruply imaged quasars,” is arXiv:2609.35974v2, revised Oct. 14, 2026. The arXiv record says it was submitted to Physical Review D. First author Daniel Gilman and co-authors argue that the data favor models containing “core-collapsed” dark matter halos, a dense end state expected in some self-interacting dark matter, or SIDM, scenarios.

That matters because cold dark matter, or CDM, is the standard cosmological picture and generally treats dark matter as effectively collisionless. SIDM instead allows dark matter particles to scatter off one another. In some models, those interactions first puff up a halo’s center into a lower-density core and later can drive it into core collapse, producing a much denser object. A confirmed population of such halos would pose a serious challenge to the usual CDM description of structure on small scales.

“Our data exhibits a strong preference for core collapsed halos in the lens model, with Bayes factors disfavoring CDM relative to SIDM ranging from 8:1 to 24:1,” the authors write in the abstract. They add: “These results pose a challenge to the CDM paradigm.”

Even so, the paper itself gives a cautionary translation of those Bayes factors into rough frequentist terms: factors of 4.3, 10 and 30 correspond to about 2.0 sigma, 2.4 sigma and 2.9 sigma, respectively. On that scale, the reported preference is suggestive, not close to the level typically associated with a discovery claim.

The study draws on 29 quadruply imaged quasars from the JWST Lensed Quasar Dark Matter Survey. A quadruply imaged quasar is a distant active galaxy whose light is bent by an intervening galaxy into four separate images. Small mismatches in the images’ brightness ratios can reveal otherwise invisible clumps of matter in and around the lensing galaxy. In this case, the analysis targets dark matter halos and subhalos in the mass range from 10^6 to 10^10.7 times the mass of the sun.

The survey uses mid-infrared, or warm-dust, flux ratios from the James Webb Space Telescope. That is important because this light is less contaminated by ordinary stellar microlensing than optical observations, making it a cleaner probe of low-mass structure.

The authors say the main systematics are the abundance of globular clusters near the lens galaxies, the abundance and tidal evolution of subhalos, and the internal structure of deeply collapsed halos. They argue that explaining the data without core collapse would require a globular-cluster abundance about five to 10 times above local-universe expectations and-or more subhalos than N-body simulations predict.

The paper also makes clear that SIDM structure-formation modeling is less well calibrated than CDM. That caveat is important: the claimed signal depends not just on the observations, but on how confidently competing dark-matter scenarios can be modeled at these scales.

To test whether their analysis pipeline might produce a spurious preference on its own, Gilman and co-authors say they ran 54 simulated survey trials assuming CDM was the true model. None produced Bayes factors above 3:1, which the authors present as evidence that their measured 8:1 to 24:1 preference is not a routine statistical fluctuation of the method.

That strengthens the case that the signal is worth attention. But for now it remains a preprint claim, not a settled result. If it survives peer review and is backed by independent follow-up, it would challenge the standard CDM picture at small scales. Until then, it is best read as an intriguing early sign that JWST’s lensed quasars may be probing new dark matter physics.