The Universe is not making stars the way it used to.
Over the past 4.5 billion years, the rate at which new stars are being formed across the cosmos has fallen by more than half. Astronomers have long tried to determine whether galaxies are simply running out of the gas needed to make them.
A new study suggests that explanation is too simple.
After combining observations covering approximately 2.5 million galaxies, researchers found that the amount of atomic hydrogen associated with galaxies has changed remarkably little over the same period.
While the cosmic star-formation rate declined by a factor of approximately 2.46, the researchers’ corrected estimate suggests that the cosmic reservoir of atomic hydrogen declined by only a factor of around 1.12.
Put differently, the Universe is producing far fewer stars even though much of the raw hydrogen from which those stars ultimately originate is still there.
The study, published on 1 September 2026 in Nature Astronomy, suggests that the slowdown in star formation is not primarily being caused by galaxies rapidly exhausting their supplies of atomic hydrogen.
Instead, the difficulty may lie in what happens to that hydrogen next.
Galaxies still have much of their raw material
Hydrogen is the most abundant element in the Universe and forms the foundation of the gas cycle that eventually produces stars.
But hydrogen does not move directly from a diffuse cloud into a star.
Gas flowing into and around galaxies passes through several stages. Much of it exists first as neutral atomic hydrogen, known to astronomers as H I. Under the right conditions, some of that atomic gas cools, becomes denser and is converted into molecular hydrogen.
It is within these cold molecular clouds that stars are ultimately born.
This creates an important distinction.
A galaxy can contain a substantial reservoir of atomic hydrogen while still forming relatively few stars if that gas is not efficiently becoming molecular.
The new research suggests that something similar may be happening across the Universe on an enormous scale.
The researchers looked back 4.5 billion years
Chuan-Peng Zhang, Hong Guo, Yizhou Gu and an international team reconstructed the changing abundance of atomic hydrogen over the period corresponding to a redshift of approximately 0 to 0.41.
In practical terms, this allowed them to compare the Universe today with conditions extending roughly 4.5 billion years into the past.
To do it, they combined two very different types of astronomical observation.
The first came from the Five-hundred-meter Aperture Spherical Telescope, better known as FAST. This enormous radio telescope can detect the characteristic radio signal emitted by neutral atomic hydrogen.
The second came from the Dark Energy Spectroscopic Instrument, or DESI, which provided optical information and precise distances for enormous numbers of galaxies.
Together, the datasets covered about 12,000 square degrees of sky and around 2.5 million galaxies.
That scale allowed the researchers to investigate the evolution of the cosmic hydrogen reservoir far more comprehensively than would be possible by examining a small number of individually bright galaxies.
Hydrogen announces itself with a radio signal
Atomic hydrogen has a particularly useful property for astronomers.
It emits radio waves at a wavelength of approximately 21 centimetres.
The signal originates when the relative orientation of the proton and electron in a hydrogen atom changes. A single event produces an almost unimaginably tiny amount of energy, but galaxies contain so many hydrogen atoms that the combined signal can be detected across enormous distances.
This 21-centimetre line has consequently become one of radio astronomy’s most important tools for mapping otherwise invisible gas.
Stars may dominate photographs of galaxies, but a substantial portion of a galaxy’s material exists in gas that optical telescopes cannot easily see.
Radio observations allow astronomers to reveal that hidden reservoir.
The expected decline was surprisingly weak
The researchers first measured what they describe as a raw decline in the cosmic density of atomic hydrogen.
Across the 4.5-billion-year period, the raw measurement decreased by a factor of 1.35 ± 0.10.
That is already substantially smaller than the decline in cosmic star formation.
The team then modelled systematic effects that could influence the measurements. Using what they describe as conservative corrections, the estimated decline became even weaker at approximately 1.12 ± 0.10.
A factor of 1.12 corresponds to only a modest reduction across billions of years.
Over the same period, however, the cosmic star-formation rate density declined by a factor of approximately 2.46.
The two trends are therefore clearly not moving together.
Galaxies did not simply use up their hydrogen
This allows the researchers to rule out one particularly straightforward explanation for the declining birth rate of stars.
If galaxies were forming fewer stars because they were rapidly consuming their atomic hydrogen reservoirs, the amount of H I should have fallen much more dramatically as star formation declined.
It did not.
The researchers also examined galaxies while accounting for their stellar mass.
At a fixed stellar mass, the average atomic-hydrogen gas fraction changed by less than 0.2 dex over the period studied.
This indicates that the weak evolution was not simply being produced by one unusual category of galaxy.
It appeared broadly across the galaxy population.
The bottleneck may come later
The result shifts attention towards the conversion between different forms of gas.
Atomic hydrogen is an important reservoir, but it is not the material from which stars form most directly.
Before large numbers of stars can develop, gas generally needs to cool and condense into molecular clouds.
Previous observations indicate that the cosmic abundance of molecular gas has declined much more closely alongside the star-formation rate.
This creates an intriguing possibility.
The Universe may not have run short of the basic ingredient for stars. Instead, galaxies may have become less effective at moving that ingredient through the next stage of the production process.
Imagine a factory with a warehouse that is still reasonably full but a production line that has slowed dramatically.
Looking only at the number of finished products could make it seem as though the factory had run out of raw materials.
The problem may actually be somewhere between storage and production.
Why would hydrogen stop becoming molecular gas?
That question is considerably more complicated.
Whether atomic gas eventually becomes a cold molecular cloud depends on its density, temperature, chemical composition and surrounding environment.
Galaxies are also not closed systems.
Gas can fall into them from the surrounding circumgalactic and intergalactic environment. It can be heated, cooled, compressed or expelled.
Stars themselves return material to their surroundings through stellar winds and supernova explosions.
Supermassive black holes can inject enormous quantities of energy into their host galaxies, potentially heating gas or preventing it from cooling efficiently enough to form new stars.
This complex circulation of material is sometimes described as the baryon cycle.
The new measurements place an important constraint on models of that cycle because any successful explanation for declining star formation now needs to preserve a comparatively large atomic-hydrogen reservoir at the same time.
A galaxy can therefore be gas-rich and quiet
The finding also illustrates why the phrase “fuel for star formation” can be misleading when applied to all hydrogen equally.
Having atomic hydrogen available does not guarantee active star formation.
The gas must reach the correct physical conditions.
If it remains too diffuse, too warm or otherwise unable to collapse into dense molecular structures, the galaxy can retain substantial amounts of hydrogen while creating relatively few new stars.
This helps explain why measuring the amount of gas present and measuring how efficiently that gas is being converted are separate scientific questions.
Our own cosmic era is relatively quiet
The decline examined in the new study is part of a much larger change in the history of the Universe.
Cosmic star formation reached its peak billions of years before the period investigated here.
The modern Universe is comparatively subdued.
Galaxies continue to form stars, including the Milky Way, but the overall rate across cosmic space is far below what it once was.
Astronomers are ultimately trying to understand why the Universe moved from a highly productive era of galaxy growth towards the quieter one we inhabit today.
The new hydrogen measurements show that at least during the most recent 4.5 billion years, simply exhausting the atomic reservoir cannot provide the full answer.
South Africa is studying the same invisible gas
This question also has a direct connection to South African astronomy.
MeerKAT, the 64-dish radio telescope in the Northern Cape, is exceptionally sensitive to the same 21-centimetre signal from neutral hydrogen.
The MeerKAT International GHz Tiered Extragalactic Exploration survey, known as MIGHTEE, specifically investigates how neutral hydrogen in galaxies evolves and how that gas eventually feeds molecular clouds and star formation.
In 2025, researchers using MeerKAT reported direct detections of neutral hydrogen in 11 galaxies at redshifts above 0.25, including one at a redshift of 0.3841.
That corresponds to a distance of more than four billion light-years.
It is particularly relevant to the new Nature Astronomy study because both programmes are pushing measurements of atomic hydrogen into an era when its faint radio signal becomes increasingly difficult to detect.
MeerKAT will eventually be incorporated into the much larger SKA-Mid telescope being constructed in South Africa.
The sensitivity of these future observations should allow astronomers to trace the cosmic gas cycle with substantially greater precision.
The measurement itself remains difficult
There are important limitations to the new result.
Detecting 21-centimetre emission becomes increasingly challenging as astronomers look farther away because the signal from individual galaxies becomes extremely faint.
Measurements of the cosmic H I density therefore depend on combining large datasets and carefully modelling observational biases.
The difference between the researchers’ raw estimate of a 1.35-fold decline and their corrected estimate of 1.12 illustrates the importance of these systematic uncertainties.
The authors deliberately describe their correction procedure as conservative, but future surveys will still be needed to test the precise magnitude of the evolution.
The study also covers the most recent 4.5 billion years rather than the entire history of star formation.
It therefore cannot by itself explain why cosmic star formation rose to its much earlier peak or everything that drove its subsequent decline.
The missing stars may tell us more about efficiency than supply
The simplest version of the story would have been easy to understand.
The Universe formed enormous numbers of stars, gradually used up its gas and now forms fewer because the fuel is disappearing.
The observations do not support that neat progression, at least not for atomic hydrogen over the period examined.
Much of the reservoir remains.
What has changed far more dramatically is the rate at which the Universe turns gas into stars.
That makes the decline of star formation a problem of conversion, regulation and environment as much as one of supply.
Future observations will need to determine where the bottleneck occurs: how gas enters galaxies, how atomic hydrogen becomes molecular, how molecular clouds collapse and how feedback from stars and black holes interferes with each stage.
The answer could help explain why the modern Universe looks the way it does.
Galaxies have not stopped forming stars because the cosmos has simply emptied its tank.
Billions of years later, much of the hydrogen is still there.
The mystery is why so much of it is no longer becoming stars.
Source Information
Study Title: Weak evolution of cosmic atomic hydrogen over the past 4.5 billion years
Lead Authors: Chuan-Peng Zhang, Hong Guo and Yizhou Gu et al.
Journal: Nature Astronomy
Published: 1 September 2026
Galaxy sample: Approximately 2.5 million galaxies
Survey area: Approximately 12,000 square degrees
Period studied: Approximately 4.5 billion years, 0 < z < 0.41
Observatories/Data: Five-hundred-meter Aperture Spherical Telescope (FAST) and Dark Energy Spectroscopic Instrument (DESI)
Raw H I decline: Factor of 1.35 ± 0.10
Systematics-corrected H I decline: Factor of 1.12 ± 0.10
Cosmic star-formation decline: Factor of 2.46
DOI: 10.1038/s41550-026-02965-9
