Largest 3D Map of the Universe Intensifies Debate Over Cosmological Expansion Rate

On April 15, 2026, the Dark Energy Spectroscopic Instrument collaboration announced it had completed its planned five-year survey of the cosmos — producing the largest high-resolution 3D map of the universe ever assembled [1]. DESI cataloged more than 47 million galaxies and quasars, far exceeding its original target of 34 million, along with 20 million stars [2]. The map spans 11 billion years of cosmic history and covers roughly one-third of the night sky [3].

The achievement is a technical triumph. But the data it has produced is making one of physics' most uncomfortable problems harder to ignore: the universe's expansion rate, as measured from nearby objects, does not match the rate predicted by observations of the ancient cosmos. That disagreement — known as the Hubble tension — now sits at 5.8 standard deviations, a level of statistical significance that in particle physics would qualify as a discovery [4]. And DESI's results are adding a second complication: evidence at 4.2 sigma that dark energy, the mysterious force driving cosmic acceleration, may not be constant at all [5].

The Numbers That Don't Add Up

The Hubble constant (H₀) describes the rate at which the universe is expanding. Two fundamentally different approaches yield two stubbornly different answers.

Measurements anchored in the early universe — particularly the European Space Agency's Planck satellite, which mapped the cosmic microwave background (CMB), the afterglow of the Big Bang — produce a value of approximately 67.4 ± 0.5 km/s/Mpc when interpreted through the standard Lambda-CDM cosmological model [4]. That model treats dark energy as a cosmological constant with a fixed equation of state.

Measurements made in the local universe tell a different story. The SH0ES collaboration (Supernova H₀ for the Equation of State), led by Nobel laureate Adam Riess at Johns Hopkins University, uses Cepheid variable stars and Type Ia supernovae as distance markers. Their most recent determination: 73.17 ± 0.86 km/s/Mpc [4]. That roughly 9% gap between the two values has persisted and, in some analyses, grown over the past decade.

DESI's baryon acoustic oscillation (BAO) measurements — which track the imprint of sound waves from the early universe frozen into the distribution of galaxies — land on the early-universe side of the divide. Combined with CMB data, DESI's second data release (DR2) yields H₀ = 68.40 ± 0.27 km/s/Mpc, a 0.4% precision measurement [6]. This is consistent with Planck but remains in sharp tension with SH0ES.

Source: Compiled from published measurements

Data as of Apr 15, 2026CSV

A Map Unlike Any Before It

To understand why these results carry weight, consider the scale of the data.

DESI's predecessor surveys from the Sloan Digital Sky Survey — BOSS and eBOSS — measured BAO distances using roughly 1.2 million and 1.5 million galaxy redshifts, respectively [7]. DESI's first data release in 2024 covered 5.7 million objects. DR2, based on three years of observations, reached 14 million [6]. The completed survey now encompasses 47 million extragalactic objects — six times more than all previous spectroscopic surveys combined [1].

Source: DESI, SDSS collaboration data

Data as of Apr 15, 2026CSV

The instrument, mounted on the Nicholas U. Mayall 4-meter Telescope at Kitt Peak National Observatory in Arizona, can simultaneously capture spectra from 5,000 objects using robotic fiber positioners [3]. Its redshift coverage extends from nearby bright galaxies (z < 0.4) through luminous red galaxies (z ~ 1.0), emission-line galaxies (z ~ 1.6), and quasars reaching beyond z = 3.5 — giving it a far broader temporal reach than BOSS or eBOSS [7].

The sheer volume of data makes certain categories of systematic error increasingly implausible as explanations for the Hubble tension. Random instrumental errors diminish as the square root of sample size. When a dataset is 30 times larger than its predecessors and produces consistent results, the probability that the discrepancy originates from instrumental noise becomes vanishingly small [1][6].

Dark Energy: Constant or Evolving?

DESI's most provocative finding concerns dark energy itself. The standard Lambda-CDM model treats dark energy as a cosmological constant — its energy density remains the same as the universe expands. DESI's BAO data, when combined with CMB and supernova observations, prefers a model in which dark energy's equation of state changes over time [5].

The collaboration fits its data to the w₀wₐCDM parameterization, where w₀ represents the present-day equation of state and wₐ captures its evolution. In DR2, the preference for evolving dark energy (w₀ > −1, wₐ < 0) over a cosmological constant reached 4.2 sigma — up from roughly 2.5 sigma in DR1 [5][8]. The combination of DESI BAO and CMB data alone shows a 3.1-sigma preference [6].

This does not yet meet the 5-sigma threshold conventionally required to claim a discovery in physics [8]. But the trend is moving in one direction: each additional year of DESI data has strengthened the signal rather than diluting it.

If confirmed, evolving dark energy would require modifying or replacing the cosmological constant — the Λ in Lambda-CDM — which has anchored standard cosmology for a quarter-century [9].

What Could Explain the Tension?

Three broad categories of theoretical explanation have attracted the most attention.

Early dark energy. These models posit a brief episode of additional dark energy in the early universe, before the CMB was emitted, that would increase the expansion rate at that epoch and shift the predicted value of H₀ upward. The idea gained traction in the early 2020s, but faces a significant constraint: CMB and large-scale structure data require that any such early component contribute less than 3% of the universe's total energy budget — well below the ~10% typically required to resolve the tension [10][11]. As a 2021 analysis in Communications Physics showed, simply reducing the cosmic sound horizon (the key mechanism by which early dark energy operates) is insufficient to fully reconcile the tension without creating new disagreements with other datasets [11].

New relativistic species. Adding extra light particles — beyond the three known neutrino species — to the early universe increases the expansion rate and can raise the inferred H₀. However, precision measurements of the CMB tightly constrain the effective number of neutrino species (Nₑff), leaving limited room for this solution [10].

Interacting dark matter. Models in which dark matter interacts with dark energy or with neutrinos can alter the expansion history. Some variants show promise, but none has yet produced a compelling fit across all available datasets without introducing new tensions elsewhere [10].

DESI's data has not decisively confirmed or ruled out any of these proposals. What it has done is tighten the constraints: the evolving dark energy signal, if it holds, would favor models in the w₀wₐ family while disfavoring pure Lambda-CDM. The full five-year dataset, expected to be analyzed in 2027, will be critical [8].

The Case for a Measurement Artifact

Not everyone is convinced the Hubble tension reflects new physics. A persistent minority view — with serious proponents — holds that unresolved systematic errors in the local distance ladder could close much of the gap.

The local measurement chain has several links, each with its own calibration uncertainties. Cepheid variable stars, whose period-luminosity relationship makes them standard candles, serve as the primary rung. But Cepheid photometry can be affected by crowding in distant galaxies, interstellar reddening, and metallicity variations [12].

The Chicago-Carnegie Hubble Program (CCHP), led by Wendy Freedman at the University of Chicago, has measured H₀ using two alternative distance indicators: the Tip of the Red Giant Branch (TRGB) and J-region Asymptotic Giant Branch (JAGB) stars. Using JWST data, the CCHP's TRGB-based measurement yields H₀ = 70.39 km/s/Mpc when combining JWST and Hubble Space Telescope observations — lower than SH0ES, though still above Planck [13]. Their JAGB-based result is lower still: H₀ = 67.80 ± 2.17 (stat) ± 1.64 (sys) km/s/Mpc, statistically consistent with the CMB value [13].

The CCHP has reported that JAGB distance moduli disagree with Cepheid distances at the 3σ level, with a weighted mean offset of +0.086 ± 0.028 magnitudes [13]. This suggests a possible systematic bias in Cepheid-based distances, though the SH0ES team has challenged this interpretation, identifying what they argue is a missing source of uncertainty in the CCHP analysis [14].

JWST observations have largely confirmed that the Hubble Space Telescope's Cepheid measurements were not significantly contaminated by crowding or dust effects — the most frequently cited instrumental concern [4]. But subtler calibration issues remain debated. A 2025 analysis in the Monthly Notices of the Royal Astronomical Society specifically revisited the Cepheid period-luminosity relation, examining the assumed prior for the residual parallax offset of Milky Way Cepheids and systematic differences in Cepheid periods between anchor galaxies and supernova host galaxies [12].

The BAO sound-horizon calibration presents a different kind of systematic question. The BAO method does not measure H₀ directly — it measures distance ratios relative to the sound horizon at recombination. Converting those ratios into an absolute H₀ requires assuming a cosmological model (typically Lambda-CDM) or combining with CMB data [6]. If that model is wrong — as the evolving dark energy signal tentatively suggests — the derived H₀ would shift. This circularity means that BAO-based H₀ values, while extremely precise, are model-dependent in ways that local measurements are not.

The Machines Built to Settle the Debate

Multiple billion-dollar missions are now operating or under construction with the explicit goal of resolving the Hubble tension and characterizing dark energy.

DESI itself will continue observing through 2028, expanding its survey area from 14,000 to 17,000 square degrees and targeting 63 million total extragalactic redshifts [1]. The project is managed by Lawrence Berkeley National Laboratory and funded primarily by the U.S. Department of Energy's Office of Science, with additional support from the National Science Foundation, the Gordon and Betty Moore Foundation, the Heising-Simons Foundation, France's CEA, Spain's Ministry of Science and Innovation, Mexico's SECIHTI, the UK's Science and Technology Facilities Council, and over 70 member institutions worldwide [15]. The collaboration includes more than 900 researchers, among them 300 PhD students [15].

Euclid, launched by the European Space Agency in July 2023, is conducting a six-year survey of the dark universe from the Sun-Earth L2 point. Its primary mission is to map the distribution of dark matter and measure the expansion history using both weak gravitational lensing and galaxy clustering [16].

The Nancy Grace Roman Space Telescope, NASA's next flagship observatory, is scheduled to launch in September 2026 at an estimated total cost of $3.9 billion [17]. Its wide-field infrared surveys will measure both supernovae and BAO, providing an independent cross-check on DESI and Euclid.

The Simons Observatory, a ground-based CMB experiment in Chile's Atacama Desert, will provide improved measurements of the CMB that can tighten constraints on early-universe physics relevant to the tension [18].

Among these, DESI's extended survey and Roman's launch are considered the near-term milestones most likely to produce decisive data. If DESI's full five-year analysis, expected in 2027, pushes the evolving dark energy signal past 5 sigma, it would constitute the first confirmed deviation from Lambda-CDM. Roman's independent BAO and supernova measurements, beginning in 2027, would either corroborate or contradict that finding [17].

Who Produces the Map, Who Interprets It

The DESI collaboration presents a structural feature common in large physics experiments but worth noting: the same team that builds the instrument and collects the data also performs the primary cosmological analysis. This is standard practice — the experimenters understand their systematics best — but it creates an environment in which the group with the strongest incentive to find interesting results is also the group making the key interpretive choices [15].

The collaboration has taken steps to mitigate this: analyses are blinded until methodology is finalized, results are reviewed internally before publication, and the data is released publicly (DESI opened its first data release in March 2025) [3]. External groups have already begun independent analyses using DESI data, and the DR2 BAO measurements are consistent with results from the Sloan Digital Sky Survey, suggesting no obvious instrumental bias [6].

Still, independent confirmation from Euclid and Roman — instruments with different systematic error profiles — will be essential before the community accepts the evolving dark energy signal as established fact.

What Breaks If the Tension Is Real

If the Hubble tension reflects genuine new physics beyond Lambda-CDM, the implications cascade through cosmology.

The cosmological constant Λ — currently treated as a fixed property of spacetime — would need to be replaced by a dynamical field. The equation of state parameter w, currently assumed to equal exactly −1, would become a function of time [9]. This would require revising textbook descriptions of the universe's energy budget, its age, and its ultimate fate.

Dark energy mission design would also shift. Current and planned surveys are optimized to distinguish between a cosmological constant and evolving alternatives. If evolving dark energy is confirmed, the next generation of experiments would need to characterize how it evolves — its functional form, its coupling to other sectors of physics, and whether it relates to the inflationary epoch [9].

The timeline for a consensus replacement model is uncertain. Lambda-CDM took roughly two decades to solidify after the 1998 discovery of accelerating expansion. A successor framework would need to accommodate not only the Hubble tension and evolving dark energy, but also the neutrino mass constraints (DESI DR2 sets an upper limit of Σmν < 0.16 eV in the w₀wₐ model [6]), the S8 tension in large-scale structure measurements, and whatever additional anomalies emerge from Euclid and Roman data.

Realistically, even with favorable data, establishing a new standard model of cosmology would require at minimum five to ten years of converging results from multiple independent experiments [9][18].

The Field Is Paying Attention

The volume of academic attention tracks with the stakes. Research output on the Hubble tension has grown dramatically: OpenAlex records show 24,696 papers published on the topic through early 2026, with 6,963 papers in 2026 alone — a 65% increase over 2025's already elevated output of 4,227 [19].

Source: OpenAlex

Data as of Jan 1, 2026CSV

This is not a niche debate. The Hubble tension sits at the intersection of observational cosmology, particle physics, and gravitational theory. Its resolution — whether through new physics, better calibration, or some combination — will shape the direction of fundamental physics for the next decade.

DESI has given the field its most precise BAO measurements and its strongest hint of evolving dark energy. What it has not yet given is a definitive answer. The full five-year dataset, combined with independent results from Euclid, Roman, and ground-based CMB experiments, will determine whether the standard model of cosmology needs a patch — or a replacement.