DESI DR2 (2025): The Evidence That Dark Energy Is Weakening

The DESI DR2 results, posted in March 2025 by the Dark Energy Spectroscopic Instrument collaboration, strengthened the hint that dark energy is weakening over time. Using more than 14 million galaxies and quasars, DESI found a 2.8 to 4.2 sigma preference for evolving dark energy over a constant, short of a 5-sigma discovery.

The question before the paper

In the standard model of cosmology, dark energy is a cosmological constant: the same density of energy in every cubic meter of space, forever. DESI’s first-year results in 2024 had found a 2.5 to 3.9 sigma hint that this is wrong, and that dark energy instead changes with time. One year of data could be a fluke. Would three years make the hint grow or fade?

What the paper found

• Lambda-CDM still fits DESI alone, but DESI’s measurements are in mild tension, 2.3 sigma, with the parameters inferred from the cosmic microwave background.
• Evolving dark energy preferred at 3.1 sigma when DESI is combined with the cosmic microwave background. The preferred model has w₀ > -1 and w_a < 0: dark energy that was stronger in the past and is weakening now.
• 2.8 to 4.2 sigma with supernovae added, depending on which supernova sample is used, for example the Dark Energy Survey’s five-year sample.
• Tighter limits on neutrinos. Assuming Lambda-CDM, the sum of neutrino masses is below 0.064 eV (95% confidence).

How they did it

DESI, on the 4-meter Mayall telescope at Kitt Peak in Arizona, uses 5,000 robotic fibers to take spectra of thousands of objects at once and build a three-dimensional map of the universe. This analysis used Data Release 2, three years of operation. From the map the team measured baryon acoustic oscillations, a fixed spacing between galaxies left by sound waves in the early universe, at many distances. Comparing its apparent size across cosmic time traces the expansion history, and so the behavior of dark energy.

Why it mattered

If confirmed, a weakening dark energy would overturn the simplest picture of dark energy as a cosmological constant that has stood since the 1998 discovery of acceleration, and it would change forecasts for the far future. A constant dark energy leads to endless, accelerating expansion. A weakening one opens other possibilities, and it pushes away from the Big Rip, which needs dark energy that grows stronger. See Big Rip vs Big Crunch vs Big Freeze and our guide to the future of the universe.

What we know now

As of October 2026, the result is a strong hint, not a discovery. A July 2026 DESI Lyman-alpha analysis, a preprint, found 2.7 sigma for DESI plus the cosmic microwave background and 3.2 sigma with supernovae, a slight move back toward a constant. The DESI survey completed its planned five-year map in April 2026 with more than 47 million galaxies and quasars; full-survey dark energy results are expected in 2027. ESA’s Euclid, described in the Euclid mission overview, and the Rubin Observatory will provide independent tests. Track it in is dark energy weakening?

Who wrote it

The paper is signed by the DESI Collaboration, which involves more than 900 researchers from over 70 institutions, with individual authors starting with M. Abdul Karim and J. Aguilar.

Read the original

DESI Collaboration, Abdul Karim, M., Aguilar, J., et al. (2025). “DESI DR2 results. II. Measurements of baryon acoustic oscillations and cosmological constraints.” Physical Review D. doi:10.1103/tr6y-kpc6. Posted to arXiv on March 18, 2025, published October 6, 2025. Free on arXiv: 2503.14738.