The universe is expanding, and physicists cannot agree on how fast. In a preprint posted in late May 2026, the H0 Distance Network Collaboration reported the most precise local measurement of the expansion rate ever achieved: 73.50 ± 0.81 kilometers per second per megaparsec, a figure precise to roughly 1.1 percent. The problem is that this number lands 7.1 standard deviations away from the value predicted by observations of the early universe. In statistical terms, that gap is enormous. It all but eliminates the possibility of a coincidence and instead suggests that something fundamental is missing from the standard model of cosmology.
A network, not a single ladder
Previous measurements of the Hubble constant relied heavily on individual techniques, each with its own vulnerabilities. The H0DN Collaboration took a different approach, weaving together three independent methods for gauging cosmic distances: Cepheid variable stars, whose brightness pulses at a rate tied to their luminosity; tip-of-the-red-giant-branch (TRGB) stars, which hit a known peak brightness before evolving further; and Type Ia supernovae, the thermonuclear explosions used as “standard candles” across vast stretches of space.
Rather than picking a winner among these techniques, the team built a covariance-weighted framework that accounts for each method’s uncertainties and the correlations between them. The result is a measurement that no single systematic error can easily derail.
Adam Riess, the Nobel laureate who co-discovered the accelerating expansion of the universe in 1998 and who leads the SH0ES team, explained the logic in a NASA blog post: the network approach matters precisely because it does not depend on any one method. If Cepheid distances carry an undetected bias, the TRGB and supernova strands act as a check, and vice versa. The SH0ES team’s own earlier Cepheid-based result, published in The Astrophysical Journal Letters in 2022, stood at 73.04 ± 1.04 km/s/Mpc. The new network value is consistent with that figure but carries a tighter error bar, which makes the conflict with early-universe predictions sharper, not softer.
The other side of the tension
On the opposite end of cosmic history sit measurements rooted in the universe’s infancy. The Dark Energy Spectroscopic Instrument’s first-year baryon acoustic oscillation (BAO) data, drawn from multiple galaxy tracers and redshift bins, provide distance markers etched into the large-scale distribution of matter. When those markers are combined with cosmic microwave background data from the Planck satellite and assumptions about Big Bang nucleosynthesis, the preferred Hubble constant falls near 67 to 68 km/s/Mpc. That range follows directly from flat Lambda-CDM, the reigning standard framework that treats the universe as geometrically flat, filled with cold dark matter, and driven apart by a cosmological constant.
The 7.1-sigma gap between that prediction and the H0DN local measurement is the starkest quantification of the so-called Hubble tension reported to date. For context, physicists typically treat a 5-sigma result as the threshold for a discovery. At 7.1 sigma, the odds of the discrepancy arising from random chance alone are vanishingly small.
Counterpoints and complications
Not every measurement lines up neatly with the local value. The Dark Energy Survey (DES) Supernova Program uses an “inverse distance ladder” that anchors 1,829 Type Ia supernovae to DESI’s BAO distances rather than to nearby stellar calibrators. That method yields a Hubble constant closer to the early-universe prediction than to the H0DN result. If both the forward and inverse ladders are internally sound, the disagreement between them raises an uncomfortable question: does the direction of the measurement, whether astronomers look outward from nearby calibrators or inward from large-scale structure, introduce biases that neither team has fully accounted for?
Astrophysical foregrounds add another layer of uncertainty. Galaxy surveys like DESI must model dust extinction, peculiar velocities, and selection effects across billions of light-years. Small systematic shifts in those corrections could, in principle, nudge the early-universe inference upward and narrow the gap. Several groups have explored this possibility, but no published analysis has demonstrated a shift large enough to close a 7-sigma discrepancy.
Meanwhile, calibration work from the Chicago-Carnegie Hubble Program, which uses JWST observations of TRGB distances, feeds into the H0DN network and provides a strand that is independent of Cepheids. The JWST data have sharpened TRGB measurements considerably. However, the full covariance matrices and simulation chains from that program have not yet been released publicly, which limits the ability of outside researchers to verify exactly how much weight the TRGB strand carries in the final network average.
The DESI collaboration has not directly addressed the 7.1-sigma figure in any public statement as of May 2026. Its cosmological parameter chains are available through public data releases, but translating those chains into a head-to-head comparison with the H0DN result requires choosing a cosmological model. Allowing the dark energy equation of state to vary, for instance, can shift the inferred Hubble constant by a few percent, meaning the severity of the tension depends partly on which theoretical lens analysts use.
Two kinds of evidence, one stubborn disagreement
Understanding the Hubble tension requires recognizing that the two sides are not measuring the same thing in the same way. The H0DN paper and the earlier SH0ES result measure distances to relatively nearby objects and convert those distances into an expansion rate. These are empirical, largely model-independent determinations. By contrast, DESI’s BAO data and the Planck cosmic microwave background constrain the expansion rate only after assuming a specific cosmological model, typically flat Lambda-CDM. The tension exists between a direct measurement and a model-dependent inference, not between two equally direct observations of the same quantity.
That distinction shapes the menu of possible resolutions. If the local measurements are wrong, the error would need to be a coordinated systematic affecting Cepheids, TRGB stars, and supernovae simultaneously. The network approach was designed to test exactly that scenario, and the H0DN team argues their data rule it out at high confidence. If the early-universe inference is wrong, the fault could lie in the cosmological model itself: perhaps an unexpected form of dark energy that changes strength over time, an extra population of light particles in the early universe, or a subtle departure from general relativity on cosmological scales. Each of these possibilities would alter how distances and expansion rates map onto the observed pattern of galaxies and the relic glow of the Big Bang.
What comes next
The field now occupies an uncomfortable position. The local distance ladder, reinforced by the H0DN network, appears internally consistent and increasingly precise. Early-universe probes, from the cosmic microwave background to baryon acoustic oscillations, also form a tightly knit picture that works remarkably well for dozens of other cosmological parameters. The fact that these two pillars disagree on the Hubble constant by more than 7 sigma forces a reckoning: either one pillar is subtly misleading, or both are accurately revealing a crack in the standard model.
Several upcoming datasets could break the stalemate. Additional JWST observations of Cepheids and TRGB stars will further tighten local calibrations and, critically, could make more of the underlying analysis publicly reproducible. Future DESI data releases and complementary surveys such as the Vera C. Rubin Observatory’s Legacy Survey of Space and Time will refine BAO measurements and stress-test the assumptions linking them to the Hubble constant. Perhaps most intriguingly, gravitational-wave “standard sirens” from merging neutron stars offer a conceptually distinct route to the expansion rate, one that bypasses both stellar calibrators and cosmological models entirely.
For now, the H0DN result raises the stakes without settling the argument. It shows that local measurements can reach percent-level precision even when they braid together multiple stellar and supernova indicators, and it confirms that the resulting Hubble constant stubbornly refuses to converge with early-universe expectations. Whether that 7.1-sigma discrepancy signals a hidden systematic or the first clean fracture in Lambda-CDM remains one of the most consequential open questions in physics. The answer, when it arrives, will reshape how we understand the history and fate of the universe itself.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
