Universe's hydrogen supply barely shrank as star formation fell by half, study finds

Astronomers have found that the universe still holds nearly as much of the raw hydrogen gas needed to build stars as it did 4.5 billion years ago, even though the rate at which new stars are born has fallen by more than half over the same span. The finding, drawn from the largest and most precise census of cosmic neutral hydrogen ever assembled, undercuts the simplest explanation for one of astronomy's most persistent puzzles: why star formation across the universe has been slowing for billions of years.

The study, published online Sept. 1 in Nature Astronomy, was led by researchers at the National Astronomical Observatories of the Chinese Academy of Sciences (NAOC), the Shanghai Astronomical Observatory and Shanghai Jiao Tong University, working with the international Dark Energy Spectroscopic Instrument (DESI) collaboration. The team combined radio observations from China's giant Five-hundred-meter Aperture Spherical radio Telescope, known as FAST, with optical spectra from roughly 2.5 million galaxies catalogued by DESI, a US Department of Energy-funded survey instrument mounted on the Mayall Telescope in Arizona.

The numbers

Cosmic star formation peaked billions of years ago and has been declining ever since; the new measurements put a precise figure on just how steep that decline has been relative to the universe's fuel supply. Roughly 4.5 billion years ago, the researchers found, the overall rate of star formation across the universe was about 2.5 times higher than it is today. Over that same stretch of cosmic time, the density of neutral atomic hydrogen — the diffuse gas that fills the space between and within galaxies — dropped by far less, to only about 1.4 times its present level.

That gap matters because neutral hydrogen, often denoted HI, is the reservoir from which star-forming galaxies ultimately draw their material. If star formation had slowed mainly because galaxies were running out of that gas, the two figures should have tracked each other closely. Instead, star formation fell roughly twice as fast as the hydrogen supply did, a mismatch the authors describe as ruling out a simple gas-exhaustion, or "fuel crisis," explanation on its own.

A signal buried in noise

Neutral hydrogen is notoriously difficult to detect at cosmological distances. Atoms of the gas emit a faint radio signal at a wavelength of 21 centimeters, produced when an electron flips its spin relative to the atom's proton, but for any single distant galaxy that signal is usually too weak to separate from background radio noise. Earlier surveys were forced to choose between depth and breadth: instruments sensitive enough to detect the 21-centimeter line in individual galaxies could only cover small patches of sky, while wide surveys lacked the sensitivity to pick out the signal at all.

The new study got around that trade-off using a technique called spectral stacking. Because DESI had already measured precise distances, or redshifts, for millions of galaxies across nearly a third of the sky, the researchers could align FAST's radio data to the expected position of the 21-centimeter signal for each galaxy and add the signals together. Individually undetectable, the combined signal from thousands of galaxies at a time became statistically robust, letting the team trace the total mass of cosmic neutral hydrogen across roughly the last third of the universe's history with a precision and sample size not previously achieved.

Who the finding reshapes

The result does not resolve the mystery of the universe's fading star-formation rate so much as redirect it. Rather than asking whether galaxies are simply running out of gas, the authors argue, the field now needs to focus on why that gas is becoming harder to convert into the dense molecular clouds where stars actually form. Stars do not condense directly out of diffuse atomic hydrogen; they form within colder, denser clouds of molecular hydrogen that atomic gas must first be compressed into. The team's leading hypothesis is that changes in how gas flows through galaxies — weakening inflows from the surrounding cosmic web and declining gas densities within galaxies themselves — have made that conversion process steadily less efficient, even as the raw atomic supply has held up relatively well.

That reframing matters most immediately to astronomers who model galaxy evolution, since it shifts attention toward the physics governing the atomic-to-molecular gas transition and away from models built primarily around gas depletion. More broadly, the work bears on a question that reaches beyond any one research specialty: the trajectory of star formation determines how many future stars, planetary systems and, eventually, potentially habitable worlds the universe will produce as it continues to age.

What researchers are saying

Jiang Peng, a researcher at NAOC and one of the study's leaders, framed the puzzle the team set out to address in a statement carried by China's state news agency Xinhua.

"Why is it becoming increasingly difficult for the universe to form new stars? This is a core question in the field of galaxy formation and evolution," said Jiang Peng, a researcher at NAOC.

Guo Hong, a researcher at the Shanghai Astronomical Observatory and a co-author of the study, said in the same statement that the persistent lack of direct observational evidence on cosmic hydrogen had been a long-standing obstacle, since the 21-centimeter signal from individual distant galaxies is so often "buried in background noises." The new approach, pairing FAST's sensitivity with DESI's sky coverage, was designed specifically to overcome that limitation. Outside coverage of the result, including a summary distributed by ScienceDaily, has similarly framed the finding as shifting the central question in the field from whether cosmic gas is disappearing to why it is increasingly resistant to forming stars.

What happens next

The authors describe the combined FAST-DESI dataset as a new observational benchmark rather than a final answer, and they say it opens a path for follow-up work rather than closing the question. Because the technique relies on statistical stacking rather than detecting individual galaxies, the next phase of research is expected to probe how the atomic-to-molecular gas conversion efficiency varies with galaxy mass, environment and cosmic epoch — testing directly whether changes in gas inflow from the surrounding cosmic web, rather than depletion, are truly the dominant driver. A preprint version of the analysis, posted to the arXiv repository ahead of formal publication, is already circulating among astronomers working on complementary molecular-gas surveys, and additional FAST observing campaigns are expected to extend the hydrogen census to fainter and more distant galaxy samples in the coming years. DESI itself continues to gather redshifts as part of its broader dark-energy survey, a dataset the Nature Astronomy authors say will keep expanding the galaxy sample available for similar stacking analyses well beyond the current study.