Study of eROSITA clusters gives first constraints on Early Dark Energy; joint fit suggests H0 ≈ 71 km/s/Mpc

A new study using galaxy clusters detected by the eROSITA X-ray telescope has delivered the first constraints on Early Dark Energy from cluster number counts, adding a new observational test to one of cosmology’s biggest disputes. The headline result is carefully qualified: the cluster data by themselves do not detect the model, but a specific joint analysis with cosmic microwave background data, CMB lensing and baryon acoustic oscillation measurements finds about a 3.2-sigma preference for nonzero Early Dark Energy and an inferred Hubble constant of 71.4 ± 1.4 kilometers per second per megaparsec.

The paper, “The SRG/eROSITA All-Sky Survey: Early dark energy and Hubble constant from the cluster mass function,” is an arXiv preprint, arXiv:2609.07834, and arXiv metadata says it has been accepted for publication in Astronomy & Astrophysics. First author Jan Strunk and 25 co-authors frame the work as the first set of Early Dark Energy constraints derived from galaxy cluster counts.

That matters because the Hubble tension — the long-running mismatch between the universe’s expansion rate inferred from the early cosmos under the standard Lambda-CDM model and the higher values found by some late-universe methods — has resisted easy explanation. Early Dark Energy is one proposed fix. In that idea, a short-lived extra energy component before or around recombination, when the cosmic microwave background was formed, changes the early-universe fit in a way that can raise the inferred Hubble constant, known as H0.

The new analysis is built on the western Galactic hemisphere portion of eROSITA’s first all-sky survey, or eRASS1, taken by the eROSITA instrument aboard the Spectrum-Roentgen-Gamma mission. The full eRASS1 cosmology sample contains 5,259 optically confirmed galaxy clusters spanning redshifts from 0.1 to 0.8. For the cosmology analysis in this paper, the researchers used the 0.1 to 0.45 subset. Because cluster-abundance cosmology depends on knowing cluster masses reliably, the team calibrated masses with overlapping weak-lensing data from the Dark Energy Survey Year 3, KiDS-1000 and HSC Year 3.

The split between the paper’s two main results is important. Using eRASS1 cluster counts alone, the authors report only an upper limit on the maximum Early Dark Energy fraction, writing in the abstract: “Using eRASS1 number counts alone, we obtain an upper limit on the maximum EDE fraction of (f_{\rm EDE} < 0.3).” In other words, eROSITA cluster counts by themselves do not provide evidence for a nonzero Early Dark Energy component.

The stronger result appears only in a combined analysis. When the eRASS1 cluster sample is fit together with ACT and Planck primary CMB data, CMB lensing measurements and DESI baryon acoustic oscillation constraints, the paper finds (f_{\rm EDE} = 0.10^{+0.04}_{-0.03}), corresponding to about a 3.2-sigma preference for nonzero Early Dark Energy, and (H_0 = 71.4 \pm 1.4) km/s/Mpc. The authors note that this multi-dataset combination excludes distance-ladder calibration data. They also argue that eRASS1’s somewhat higher inferred S8 — a measure tied to how matter clusters in the universe — helps pull the combined fit toward Early Dark Energy.

That makes the study notable, but not decisive. Recent CMB-based analyses have often found weaker support for Early Dark Energy or only upper limits, so the eROSITA result stands out mainly because it introduces an independent probe rather than because it settles the case. It also comes with a familiar caution: cluster-count cosmology is especially sensitive to mass calibration, which is why the weak-lensing work from DES, KiDS and HSC is central to the analysis and likely to draw close scrutiny.

Even with those caveats, the paper marks a milestone for eROSITA’s cosmology program. A mission built in part to assemble huge, uniformly selected galaxy-cluster samples has now produced what the authors describe as the first Early Dark Energy constraints from cluster number counts — a new entry point into the Hubble tension debate, with the strongest signal appearing only in one particular joint dataset combination.