DESI's Lyman-Alpha Forest Keeps Evolving Dark Energy in Play
A full-shape analysis of DESI quasar light measures early-universe geometry to 1%. Evolving dark energy is still preferred, below discovery level.
New results from the Dark Energy Spectroscopic Instrument (DESI) give its sharpest constraints yet on the early universe from the Lyman-alpha forest, and they keep the idea of an evolving dark energy alive without settling it. The DESI collaboration announced the analysis on July 30, 2026, with a preprint posted to arXiv that is not yet peer reviewed.
The Lyman-alpha forest is a dense pattern of absorption lines that hydrogen gas imprints on the light of distant quasars. This analysis uses quasar light emitted about 11 billion years ago, from DESI’s second data release (DR2). Instead of using only the baryon acoustic oscillation (BAO) peak, a standard ruler left by sound waves in the early universe and first detected in 2005, the team fit the full shape of the signal. That measures a geometric distortion called the Alcock-Paczynski effect to 1% precision at an effective redshift of 2.33, twice as tight as BAO alone.
Assuming the standard LambdaCDM model, the data give a Hubble constant of 66.5 km/s/Mpc (give or take 1.3) and a matter fraction of 0.325 (give or take 0.018). The new value shifts slightly toward LambdaCDM, shrinking the mismatch between DESI and the cosmic microwave background from 2.4 to 2.2 sigma. An evolving dark energy model is still preferred at 2.7 sigma for DESI plus CMB, and 3.2 sigma with supernovae, short of the 5 sigma discovery threshold. That echoes the 2025 DR2 BAO results.
Whether dark energy is constant or weakening decides the long-term fate of the universe. More in our dark energy guide.
What to watch next: DESI has finished its planned survey, and first dark energy results from the full five-year data are expected in 2027.
Primary source
This summary is based on DESI Collaboration and the peer-reviewed paper. Read the original for full details, data and caveats.
This week in papers
The week's most important results, one line each, every week.
