The dynamics of the universe depend on a delicate balance between gravitational attraction from matter and the repulsive effect of dark energy. A universe containing only matter would eventually slow down its expansion due to gravitational forces and possibly recollapse. However, observations of Type Ia supernovae in the late 1990s revealed that our universe’s expansion is in fact accelerating, requiring the introduction of dark energy. The standard cosmological model, called the Lambda Cold Dark Matter (ΛCDM) model, provides an elegant and robust explanation of cosmological observations by including normal matter, cold dark matter (CDM) and dark energy. It is the foundation of our current understanding of the universe.
Cosmological constant
In ΛCDM, Λ refers to the cosmological constant – a parameter introduced by Albert Einstein to counter the effect of gravity in his pursuit of a static universe. With the knowledge that the universe is accelerating, Λ is now used to quantify this acceleration. An important parameter that describes dark energy, and therefore influences the evolution of the universe, is its equation-of-state parameter, w. This value relates the pressure dark energy exerts on the universe, p, to its energy density, ρ, via p = wρ. Within ΛCDM, w is –1 and ρ is constant – a combination that has to date explained observations well. However, new results by the Dark Energy Spectroscopic Instrument (DESI) put these assumptions under increasing stress.
These new results are part of the second data release (DR2) from DESI. Mounted on the Nicholas U Mayall 4-metre telescope at Kitt Peak National Observatory in
Arizona, DESI is optimised to measure the spectra of a large number of objects in the sky simultaneously. Joint observations are possible thanks to 5000 optical fibres controlled through robots, which continuously optimise the focal plane of the detector. Combined with a highly efficient processing pipeline, this allows DESI to perform detailed simultaneous spectrometer measurements of a large number of objects in the sky, resulting in a catalogue of measurements of the distance of objects based on their velocity-induced shift in wavelength, or redshift. For its first data release, DESI used 6 million such redshifts, allowing it to show that w was several sigma away from its expected value of –1 (CERN Courier May/June 2024 p11). For DR2, 14 million measurements are used, enough to provide strong hints of w changing with time.
The first studies of the expansion rate of the universe were based on redshift measurements of local objects, such as supernovae. As the objects are relatively close, they provide data on the acceleration at small redshifts. An alternative method is to use the cosmic microwave background (CMB), which allows for measurements of the evolution of the early universe through complex imprints left on the current distribution of the CMB. The significantly smaller expansion rate measured through the CMB compared to local measurements resulted in a “Hubble tension”, prompting novel measurements to resolve or explain the observed difference (CERN Courier March/April 2025 p28). One such attempt comes from DESI, which aims to provide a detailed 3D map of the universe focusing on the distance between galaxies to measure the expansion (see “3D map” figure).

The 3D map produced by DESI can be used to study the evolution of the universe as it holds imprints from small fluctuations in the density of the early universe. These density fluctuations have been studied through their imprint on the CMB, however, they also left imprints in the distribution of baryonic matter until the age of recombination occurred. The variations in baryonic density grew over time into the varying densities of galaxies and other large-scale structures that are observed today.
The regions originally containing higher baryon densities are now those with larger densities of galaxies. Exactly how the matter-density fluctuations evolved into variations in galaxy densities throughout the universe depends on a range of parameters from the ΛCDM model, including w. The detailed map of the universe produced by DESI, which contains a range of objects with redshifts up to 2.5, can therefore be fitted against the ΛCDM model.
Among other studies, the latest data from DESI was combined with that of CMB observations and fitted to the ΛCDM model. This worked relatively well, although it requires a lower matter-density parameter than found from CMB data alone. However, using the resulting cosmological parameters results in a poor match with the data for the early universe coming from supernova measurements. Similarly, fitting the ΛCDM model using the supernova data results in poor agreement with both the DESI and CMB data, thereby putting some strain on the ΛCDM model. Things don’t get significantly better when adding some freedom in these analyses by allowing w to differ from –1.
The new data release provides significant evidence of a deviation from the ΛCDM model
An adaption of the ΛCDM model that results in an agreement with all three datasets requires w to evolve with redshift, or time. The implications for the acceleration of the universe based on these results are shown in the “Tension with ΛCDM” figure, which shows the deceleration rate of the expansion of the universe as a function of redshift. q < 0 implies an accelerating universe. In the ΛCDM model, acceleration increases with time, as redshift approaches 0. DESI data suggests that the acceleration of the universe started earlier, but is currently less than that predicted by ΛCDM.
Although this model matches the data well, a theoretical explanation is difficult. In particular, the data implies that w(z) was below –1, which translates into an energy density that increases with the expansion; however, the energy density seems to have peaked at a redshift of 0.45 and is now decreasing.
Overall, the new data release provides significant evidence of a deviation from the ΛCDM model. The exact significance depends on the specific analysis and which data sets are combined, however, all such studies provide similar results. As no 5σ discrepancy is found yet, there is no reason to discard ΛCDM, though this could change with another two years of DESI data coming up, along with data from the European Euclid mission, Vera C Rubin Observatory, and the Nancy Grace Roman Space Telescope. Each will provide new insights into the expansion for various redshift periods.