The Hubble tension is a persistent disagreement between two ways of measuring how fast the universe is expanding. One method, which studies light from the early universe, puts the expansion rate at about 67 km/s/Mpc. The other, which measures distances to relatively nearby objects, lands closer to 73 km/s/Mpc. That gap of roughly 8% has stubbornly refused to close despite years of increasingly precise measurements, and it has grown statistically significant enough that many cosmologists suspect it points to something genuinely missing from our understanding of the universe.
Two Numbers That Should Agree but Don’t
The expansion rate of the universe is captured by a single number called the Hubble constant, usually written as Hâ‚€. Think of it as a speedometer for the cosmos: for every megaparsec of distance (about 3.26 million light-years), galaxies are moving apart at Hâ‚€ kilometers per second. If our best model of the universe is correct, you should get the same answer for Hâ‚€ whether you measure it by looking at ancient light from the cosmic microwave background (the faint afterglow of the Big Bang) or by measuring distances to exploding stars in galaxies much closer to us.
The cosmic microwave background approach, anchored by the European Space Agency’s Planck satellite, gives a value near 67 km/s/Mpc. A maximum entropy analysis of Planck’s temperature data arrived at a mean of about 67 km/s/Mpc, consistent with the standard cosmological model’s prediction.1PubMed Central. Maximum Entropy Estimates of Hubble Constant from Planck Measurements Baryon acoustic oscillation measurements, which track a different imprint of the early universe’s sound waves frozen into the distribution of galaxies, independently land at a similar value of about 67.6 km/s/Mpc.2Journal of Cosmology and Astroparticle Physics. Baryon Acoustic Oscillations and the Hubble Constant: Past, Present and Future
The local, or “late-universe,” measurement uses a technique called the distance ladder. You calibrate the brightness of Cepheid variable stars in nearby galaxies, use those to figure out how far away certain supernovae are, and then watch how fast those supernova-hosting galaxies are receding. The SH0ES collaboration, the most prominent group doing this work, reports a value of about 73.2 km/s/Mpc. Surveys summarizing a wide range of such measurements conclude that the discrepancy between early-universe and late-universe values sits between 4 and 6 standard deviations, well past the threshold physicists normally use to call something real.3Classical and Quantum Gravity. In the realm of the Hubble tension—a review of solutions
Could Someone Just Be Measuring It Wrong?
The most prosaic explanation would be that one side has an overlooked systematic error: maybe the Cepheid stars used as distance markers are being measured slightly wrong, or maybe the cosmic microwave background analysis depends on assumptions that are subtly off. This has been the subject of intense scrutiny, and the short version is that neither side’s errors seem large enough to close the gap on their own.
One major concern about the distance ladder was that Hubble Space Telescope images might be blending the light of Cepheid stars with nearby unresolved stars, making the Cepheids look brighter (and therefore closer) than they really are. If this “crowding” effect worsened with distance, it could inflate the measured expansion rate. The James Webb Space Telescope, with its sharper infrared vision, provided a direct test. By comparing over a thousand Cepheids observed by both Hubble and JWST in the same galaxies, the SH0ES team found a mean distance difference of essentially zero and rejected distance-dependent crowding as an explanation for the tension at very high confidence.4The Astrophysical Journal Letters. JWST Observations Reject Unrecognized Crowding of Cepheid Photometry as an Explanation for the Hubble Tension at 8σ Confidence This was arguably the single most important observational development in recent years: the worry that Hubble’s aging optics were fooling us has been largely put to rest.
Other systematics in the Cepheid distance ladder remain under discussion. The brightness of a Cepheid depends somewhat on its chemical composition, and different teams get different estimates for how strong that effect is. Reddening by interstellar dust, cross-calibration between instruments, and sample selection all introduce their own small uncertainties.5Research in Astronomy and Astrophysics. Distance-ladder Measurements of the Hubble Constant: Recent Progress, Systematics, and Prospect One recent reanalysis that applied a different treatment of Cepheid calibration data reported reducing the tension from 5.4 standard deviations down to 2.4, suggesting that the way Cepheid data is processed still matters a lot.6arXiv. Reassessing the Cepheid-based distance ladder: implications for the Hubble constant That finding is contested, though, and it highlights how sensitive the final number is to seemingly technical choices in the analysis pipeline.
What Other Measurement Methods Say
If both the cosmic microwave background and the Cepheid-supernova distance ladder have been checked thoroughly, the next natural question is whether independent methods land on one side or the other. Several alternatives exist, and they paint a complicated picture.
The tip of the red giant branch (TRGB) method uses a different kind of star as a distance marker. When old, low-mass red giant stars reach a critical brightness just before a dramatic internal change, they all hit roughly the same peak luminosity, making them useful yardsticks. Different teams using TRGB have gotten somewhat different answers depending on how they calibrate the method. The Carnegie-Chicago Hubble Program initially reported a value of about 69.8 km/s/Mpc, which sat comfortably between the two camps.7The Astrophysical Journal. The Carnegie-Chicago Hubble Program. VIII. An Independent Determination of the Hubble Constant Based on the Tip of the Red Giant Branch But an independent reduction of the same program’s data, using the same TRGB technique with different analysis choices and new Hubble observations, found about 71.5 km/s/Mpc, closer to the distance-ladder camp.8The Astrophysical Journal. Comparing Tip of the Red Giant Branch Distance Scales: An Independent Reduction of the Carnegie-Chicago Hubble Program and the Value of the Hubble Constant A calibration using the Gaia satellite’s precise measurement of the distance to the star cluster Omega Centauri yielded about 72.1 km/s/Mpc.9The Astrophysical Journal Letters. The Parallax of ω Centauri Measured from Gaia EDR3 and a Direct, Geometric Calibration of the Tip of the Red Giant Branch and the Hubble Constant
Meanwhile, JWST has now measured TRGB distances directly and compared them with Cepheid distances for the same galaxies, finding no meaningful difference between the two methods.10The Astrophysical Journal. Tip of the Red Giant Branch Distances with JWST. II. I-band Measurements in a Sample of Hosts of 10 Type Ia Supernova Match HST Cepheids The TRGB results, in other words, have been drifting toward the higher end rather than splitting the difference.
Water megamasers offer yet another approach. Certain galaxies have clouds of water vapor orbiting their central black holes that act like natural lasers in the microwave spectrum. By mapping these masers’ positions and velocities with radio telescopes, astronomers can get a purely geometric distance without relying on any kind of standard candle. Results from individual megamaser galaxies have produced values such as about 69 km/s/Mpc for UGC 3789 and about 66 km/s/Mpc for NGC 5765b, both with uncertainties around 10%.11The Astrophysical Journal. THE MEGAMASER COSMOLOGY PROJECT. IV. A DIRECT MEASUREMENT OF THE HUBBLE CONSTANT FROM UGC 378912The Astrophysical Journal. THE MEGAMASER COSMOLOGY PROJECT. VIII. A GEOMETRIC DISTANCE TO NGC 5765b These are consistent with either camp, given their wide error bars, but they are genuinely independent of the usual stellar distance indicators.
Gravitational Lensing and Gravitational Waves
Two newer methods promise to break the deadlock because they sidestep both the distance ladder and early-universe physics entirely.
Time-delay cosmography exploits strong gravitational lensing, where a massive galaxy or cluster bends light from a distant quasar into multiple images. Because each image’s light travels a different path, the images arrive at slightly different times. That time delay depends on the actual distances involved, which in turn depend on Hâ‚€. The H0LiCOW and TDCOSMO collaborations have used this approach, and early results tend to favor the higher end, though the uncertainties are still large enough to be interesting rather than decisive. With upcoming survey telescopes, this technique is expected to reach 1% precision and could provide a genuinely independent verdict.13PubMed Central. Time-Delay Cosmography: Measuring the Hubble Constant and Other Cosmological Parameters with Strong Gravitational Lensing
Gravitational waves offer a completely different kind of measurement, sometimes called “standard sirens.” When two compact objects such as neutron stars or black holes merge, the gravitational waves they emit encode information about their distance. If you can also identify the host galaxy (either through electromagnetic light from the merger or by statistically cross-matching with galaxy catalogs), you can measure both the distance and the recession speed independently. A measurement using the black-hole merger GW170814 and galaxy catalog data from the Dark Energy Survey demonstrated the method’s feasibility, though individual events still carry large uncertainties.14The Astrophysical Journal Letters. First Measurement of the Hubble Constant from a Dark Standard Siren using the Dark Energy Survey Galaxies and the LIGO/Virgo Binary–Black-hole Merger GW170814 Accumulating dozens or hundreds of such events over the coming decade could narrow the uncertainty to the point where gravitational waves settle the question on their own. A recent blinded mock data challenge, however, warned that if the mass distribution of merging black holes is not independently well understood at all relevant distances, the technique could return a biased answer for Hâ‚€.
What If the Universe’s Rulebook Is Incomplete?
If neither side’s measurements are wrong, the gap may be telling us that our standard model of cosmology, known as ΛCDM (for cold dark matter plus a cosmological constant representing dark energy), is missing something. Theorists have proposed a strikingly long menu of possible fixes, numbering in the hundreds. Most fall into two broad categories depending on when they modify the universe’s behavior.
Early-universe proposals tinker with physics before the cosmic microwave background was emitted, roughly 380,000 years after the Big Bang. The most studied of these is early dark energy: a brief pulse of extra energy that acts like a cosmological constant at very early times and then fades away quickly. This extra energy speeds up the early expansion just enough to shrink the “sound horizon,” a characteristic distance scale imprinted in the cosmic microwave background. A smaller sound horizon, when analyzed assuming ΛCDM, naturally yields a higher inferred value of Hâ‚€, potentially closing the gap.15PubMed. Early Dark Energy can Resolve the Hubble Tension Variants of this idea, such as “New Early Dark Energy” driven by a first-order phase transition just before recombination, attempt to accomplish the same shift with different underlying physics.16Physical Review D. Resolving the Hubble Tension with New Early Dark Energy The challenge for early dark energy models is that they need to appear and vanish in a very narrow window of cosmic history without disturbing the other exquisitely well-measured features of the cosmic microwave background.
Late-universe proposals, by contrast, modify the behavior of dark energy or gravity at more recent cosmic times. Some introduce a dark energy that changes over time instead of remaining constant, allowing the expansion rate to be higher today than ΛCDM predicts. One approach combines a modified theory of gravity with a dynamical dark energy component that transitions from an exotic “phantom” phase in the early universe to a more standard behavior at low redshifts.17Monthly Notices of the Royal Astronomical Society: Letters. Metric f(R) gravity with dynamical dark energy as a scenario for the Hubble tension Another group, fitting a parameterized dark energy model to recent supernova and spectroscopic survey data, recovered a best-fit Hâ‚€ of about 73.3 km/s/Mpc, consistent with the local distance-ladder value.18New Astronomy. A parameterized equation of state for dark energy and Hubble tension The recurring difficulty for late-universe models is that they can fit the expansion data but often struggle when confronted with other measurements, such as how fast cosmic structures grow over time. Modified gravity scenarios have been shown to alleviate the Hubble tension in some data combinations while offering minimal improvement in overall fit compared to ΛCDM.
Extra Particles and Exotic Neutrinos
A different family of proposals changes the contents of the early universe rather than the behavior of dark energy. The cosmic microwave background’s pattern is sensitive to how many species of lightweight particles were zipping around at nearly the speed of light in the first few minutes. Adding extra “dark radiation” speeds up early expansion and can shift the inferred Hâ‚€ upward. With current data plus the SH0ES measurement, the preferred number of such relativistic species comes out slightly higher than the standard three neutrino types, reducing the Hubble tension to about 2.4 standard deviations. The best-fit scenario in one analysis favors a partially thermalized sterile neutrino with a mass around 0.1 to 0.2 electron-volts.19Physical Review D. Extra radiation cosmologies: Implications of the Hubble tension for eV-scale neutrinos
The problem is that such extra neutrinos would need to interact with ordinary matter in a very specific way. Large, nonstandard neutrino self-interactions were proposed as another mechanism to shift Hâ‚€ upward, but the force carrier responsible would need a large coupling and a mass in the keV to 100 MeV range. Almost all realizations of such a particle are already ruled out by existing data, unless the mediator carries spin 0 and couples almost exclusively to tau-flavored neutrinos, which is a very narrow escape hatch.20PubMed. Constraining the Self-Interacting Neutrino Interpretation of the Hubble Tension The broader lesson here is that new-physics solutions tend to solve one problem while creating new tensions elsewhere, and the ones that survive all existing constraints are increasingly contrived.
Could We Simply Live in an Unusual Patch of the Universe?
A simpler-sounding idea is that our local neighborhood is slightly less dense than the cosmic average: a “local void” that would make nearby galaxies appear to recede faster than the universal expansion rate. If we happen to sit inside such a void, distance-ladder measurements would overestimate Hâ‚€ while the cosmic microwave background, which averages over the entire visible universe, would give the true value.
This hypothesis has been tested by fitting a model where the local universe has a lower-than-average density (technically a Lemaître-Tolman-Bondi void embedded in a standard cosmology) to supernova data. The result is telling: if you restrict the analysis to the narrow redshift range used to measure Hâ‚€ locally (roughly 0.023 to 0.15), the void model works and the tension disappears. But when you use the full supernova sample extending to much greater distances, the void’s explanatory power vanishes.21Classical and Quantum Gravity. A void in the Hubble tension? The end of the line for the Hubble bubble This effectively rules out the simplest version of the “Hubble bubble” scenario, though more complex inhomogeneous cosmologies remain in play at the margins.
The S₈ Tension and Whether the Problems Are Connected
The Hubble tension is not the only crack in the standard model. A separate discrepancy, called the S₈ tension, involves how clumpy matter is in the universe today. The cosmic microwave background predicts a certain level of matter clustering, and low-redshift galaxy surveys consistently find slightly less clumping than predicted. The gap is milder than the Hubble tension but points in a suggestive direction: both tensions involve the cosmic microwave background predicting one thing and later measurements finding another.22Physical Review D. Late Time Modification of Structure Growth and the S8 Tension
Whether the two tensions share a common origin is an open and fascinating question. Some theoretical models that increase Hâ‚€ also change the predicted growth of cosmic structure, potentially addressing both at once. Others fix one while worsening the other. No single model has convincingly resolved both simultaneously while remaining consistent with all other cosmological data, which makes the search for such a model one of the more active areas in theoretical cosmology.
What Comes Next
Several observational programs in the next five to ten years are positioned to either sharpen the tension into an undeniable signal of new physics or reveal a subtle systematic that finally closes the gap. The Dark Energy Spectroscopic Instrument (DESI) is already collecting data on millions of galaxies and quasars, mapping baryon acoustic oscillations with unprecedented precision across cosmic time. The Vera Rubin Observatory’s Legacy Survey of Space and Time (LSST) will discover thousands of new Type Ia supernovae and gravitationally lensed systems. The European Space Agency’s Euclid mission is mapping the geometry of the universe with wide-field infrared surveys. Gravitational-wave detectors, both the current LIGO-Virgo-KAGRA network and planned next-generation facilities, will accumulate enough standard siren events to provide a wholly independent Hâ‚€ measurement.
The convergence of these programs matters because the Hubble tension has resisted resolution partly because each individual method carries its own idiosyncratic systematics. If four or five independent methods all point to a high Hâ‚€, the case for new physics becomes overwhelming. If one or two methods drift toward the low value as their systematics are better understood, the tension could dissolve. Right now, the trend has been for new and independent measurements to mostly land on the higher side, but with error bars large enough to maintain suspense. The field is in an unusual position: the data are precise enough to be deeply uncomfortable for the standard model, but not yet precise enough to tell us what should replace it.