For more than a decade, physicists hunting dark matter in the Milky Way’s core kept circling the same clue: a faint glow of gamma rays peaking around 2 GeV, maddeningly ambiguous, never quite proving that invisible particles were annihilating at the galaxy’s heart. Now a fresh analysis of 15 years of data from NASA’s Fermi Gamma-ray Space Telescope has surfaced something different. A team of researchers reported in a preprint posted to arXiv in July 2025 that they detected a roughly spherical halo of gamma-ray emission centered on the Galactic core, with its spectrum peaking near 20 GeV, about ten times the energy of the older signal. If the finding survives independent scrutiny, it could be the strongest gamma-ray evidence yet that dark matter particles are destroying each other at the center of our galaxy.
The instrument behind the claim
The technical backbone here is solid and publicly documented. Fermi’s Large Area Telescope (LAT) has orbited Earth since 2008, scanning the entire sky in gamma rays every three hours. In 2015, the mission rolled out a major data-processing overhaul called Pass 8, which rebuilt how the instrument classifies incoming photons, rejects cosmic-ray contamination, and reconstructs particle directions. The upgrade substantially improved sensitivity at higher energies, and that improvement is what makes a 15-year stacked search at 20 GeV feasible in a way it simply was not before.
To hunt for an unexpected signal, analysts must first subtract every known source of gamma-ray light. The standard toolkit for that job includes NASA’s Galactic interstellar emission model (gll_iem_v07.fits) and a family of isotropic spectral templates (iso_P8R3_*), both maintained on the Fermi Science Support Center’s background models page. These files encode the expected glow from cosmic rays slamming into interstellar gas and dust, plus a uniform wash from unresolved extragalactic sources. Anything left over after those templates are peeled away demands an explanation.
How this differs from the classic Galactic-center excess
The Milky Way’s center has been a contested crime scene in gamma-ray astronomy since roughly 2009, when early Fermi analyses revealed a surplus of 1-to-3 GeV photons radiating from the inner galaxy. That signal, often called the Galactic-center excess (GCE), sparked hundreds of papers debating whether it came from dark matter annihilation or from a swarm of thousands of unresolved millisecond pulsars. The Fermi-LAT Collaboration’s own systematic study of the GCE stressed that uncertainties in cosmic-ray source distributions, interstellar gas maps, and Fermi-bubble modeling can mimic or absorb a dark matter signal, leaving the question unresolved.
The newly reported halo is a different beast. Its energy is roughly an order of magnitude higher, peaking near 20 GeV rather than 2 GeV. According to the preprint, the excess persists after the authors tested it against cataloged point sources, GALPROP-based diffuse emission models, the isotropic template, Loop I emission, and the Fermi bubbles. The authors argue that its approximately spherical morphology and spectral shape fit expectations for dark matter annihilation better than known astrophysical processes.
Whether the two signals represent separate phenomena, or whether one is an artifact of how the other is modeled, has not been resolved. That ambiguity matters: if the background model used to subtract the GCE is slightly wrong, it could create or distort a residual at higher energies.
Why caution is warranted
Several important caveats apply as of June 2026. The preprint has not yet passed formal peer review. The Fermi-LAT Collaboration has not issued a public statement endorsing or independently replicating the result. The custom likelihood analysis code and full residual maps used to isolate the 20 GeV component have not been publicly released beyond the arXiv paper itself.
No cross-checks against data from other gamma-ray instruments have been reported. Ground-based Cherenkov telescopes such as H.E.S.S., which has published observations of the Galactic center in overlapping energy ranges, could provide an external consistency test. The upcoming Cherenkov Telescope Array (CTA), designed to be far more sensitive at tens of GeV and above, would be an even more powerful arbiter, but its full southern-array operations are still ramping up.
Background modeling remains the central vulnerability. The Fermi-LAT Collaboration’s earlier work on the 1-to-3 GeV excess illustrates the problem vividly: that signal has persisted for years, yet plausible variations in cosmic-ray propagation models and gas maps can shift it substantially. The 20 GeV halo faces an analogous challenge at higher energies, where the telescope’s effective area is smaller and each photon carries more statistical weight. Small mismodeling errors in the diffuse background can masquerade as a new component or erase a real one.
What would make or break the signal
The most immediate test is straightforward replication. Because the analysis relies on standard Pass 8 event selections and widely used diffuse models, independent teams should be able to reproduce the basic residual maps using publicly available Fermi data. If they find a similar spherical component peaking around 20 GeV, the conversation shifts from “Is it real?” to “What causes it?” If modest tweaks to the diffuse emission model erase the halo, the result will likely join a long list of intriguing but transient gamma-ray anomalies.
A more targeted check involves the halo’s spatial profile. If the 20 GeV excess genuinely arises from dark matter annihilation, its brightness should flatten at higher Galactic latitudes in a characteristic pattern dictated by the dark matter density distribution. That flattening could, in principle, be tested with existing Fermi data and the publicly available gll_iem_v07 template, no proprietary tools required. The fact that such a test is accessible makes the claim falsifiable on a relatively short timescale.
Independent groups can also stress-test the result by swapping in alternative diffuse emission models, varying energy cuts, or masking different sets of point sources. Even a non-detection at comparable energies from H.E.S.S. or early CTA data would constrain how bright the halo can be and still evade other instruments.
Where dark matter hunting goes from here
For theorists, the reported spectrum and morphology offer concrete targets. Models of dark matter particles in the tens-of-GeV to TeV mass range can be tuned to produce a 20 GeV gamma-ray peak, but they must also satisfy constraints from other observations, including measurements of the cosmic microwave background and searches in dwarf satellite galaxies, where dark matter signals would be cleaner but fainter.
Astrophysical explanations have not been ruled out. An unresolved population of pulsars, or past outbursts from the supermassive black hole Sagittarius A*, could in principle produce a high-energy halo. But those scenarios would need to explain why the emission is approximately spherical and why it does not conflict with existing radio and X-ray observations of the same region.
Whatever the outcome, the episode highlights how much science remains buried in long-baseline space telescope archives. The same Pass 8 upgrade that enabled this analysis has already extended Fermi’s reach into fainter structures and higher energies. As more researchers mine those 15-plus years of photon data with fresh statistical techniques, additional surprises are likely, whether or not they involve dark matter. For now, the 20 GeV halo stands as the most energetic and morphologically distinct gamma-ray anomaly reported from the Galactic center since the original GeV-scale excess. Its fate will depend on how quickly the broader community can probe, stress-test, and either confirm or refute the signal.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
