On September 1, 2026, *Nature Astronomy* published a paper by an international collaboration including the National Astronomical Observatories of the Chinese Academy of Sciences (NAOC), Shanghai Astronomical Observatory, and Shanghai Jiao Tong University. Using combined data from China's FAST radio telescope and the Dark Energy Spectroscopic Instrument (DESI), the team performed a high-precision measurement of cosmic neutral hydrogen evolution over the past 4.5 billion years. The conclusion overturns a long-standing assumption: the star formation rate has dropped to about 40% of its value 4.5 billion years ago, yet the neutral hydrogen reservoir still holds more than 71% of its former level. The fuel hasn't run out — the conversion chain is the bottleneck.
The Opening Question
Why does the universe increasingly struggle to form new stars?
The most intuitive explanation in astronomy: stars form when cold gas (mainly neutral hydrogen) collapses; as star formation consumes the cold gas, star formation inevitably declines. If this picture were correct, then as the star formation rate falls, the cold gas reservoir should decline in lockstep.
For decades, instruments were not sensitive enough to test this. The 21 cm line of neutral hydrogen is extremely faint — a single distant galaxy is nearly buried in noise. Dr. Zhang Chuanpeng of NAOC offered an analogy: "It's like trying to hear someone whispering in a noisy plaza."
This time, FAST's high sensitivity combined with DESI's massive spectroscopic survey satisfied both depth and breadth simultaneously for the first time.
How to Extract the "Breathing of Neutral Hydrogen" from Noise
The observing strategy boils down to three steps:
1. Large sample: align the radio signals of about 2.5 million galaxies across roughly one-third of the sky by redshift; 2. Spectral stacking: each galaxy's signal is buried in noise, but stacking 2.5 million weak signals is like millions of people whispering at once in a plaza — a statistically meaningful "average signal" emerges; 3. Cross-matching with DESI spectra: DESI provides precise redshifts and galaxy properties, so the FAST H I data are stacked in bins at those redshifts.
This method says nothing about individual galaxies (they cannot be seen), but it is powerful for average values and evolutionary trends — a classic stacking analysis in which collective statistics replace single-object breakthroughs.
Key Numbers: An "Asymmetric Decline" Over 4.5 Billion Years
| Dimension | 4.5 Gyr ago | Today | |---|---|---| | Cosmic star formation rate | ≈ 2.5× today | 1.0 (baseline) | | Neutral atomic hydrogen (H I) density | ≈ 1.4× today | 1.0 (baseline) | | Ratio (H I / SFR) | 1.4 / 2.5 ≈ 0.56 | 1.00 |
The star formation rate has fallen to roughly 40% of its former value, but the neutral hydrogen reservoir has only declined to about 71%. The two curves do not decline in lockstep, directly refuting the hypothesis that fuel exhaustion is the main driver of declining star formation.
The Real Story: The Fuel Remains, but the "Factory" Has Slowed
Stars do not form directly from neutral hydrogen. H I must first be converted into denser molecular hydrogen clouds (molecular clouds), which are the true "nurseries" of stars.
Dr. Zhang Chuanpeng's causal chain:
> "What has actually changed in the late universe may be the gas conversion efficiency — as cosmic-web gas supply weakens and gas densities drop, the efficiency of converting neutral hydrogen into molecular hydrogen declines. The star-forming molecular gas gradually decreases, while neutral hydrogen is instead preserved."
Researcher Guo Hong summarized it more colloquially: "What we're seeing is not a universe running out of gas, but gas that is increasingly hard to 'process' into stars."
Context: Astronomy in Late 2026
This result fits into a broader pattern of recent work rewriting the physical foundations of galaxy evolution:
- Aug 26 — Purple Mountain Observatory's Milky Way H I survey and an intermediate-mass black hole candidate;
- Aug 26 — stacking analysis of 217 "little red dots" showing extended host-galaxy emission;
- Aug 29 — JWST's 8.7-hour NIRSpec mapping of NGC 4696, revealing a supermassive black hole self-feeding via cold gas filaments;
- Aug 31 — RHIC STAR results challenging the 53-year-old quark-sharing hypothesis for baryon number;
- Sep 1 — IRIS observations of 3D magnetic reconnection in solar flare ribbon bright kernels;
- Sep 1 — FAST × DESI: neutral hydrogen does not decline in step with star formation; the gas conversion chain is the bottleneck.
FAST's Tenth Anniversary: From Catching Up to Defining Observational Benchmarks
FAST, completed on September 25, 2016, turns ten in 2026. This paper applies its capability in a new dimension — large-scale evolution. Deputy Director Jiang Peng reflected:
> "From chasing pulsars and capturing fast radio bursts to mapping the evolution of cosmic neutral hydrogen, in the decade since completion, the Chinese Sky Eye rooted deep in the mountains has continuously produced world-class scientific results, driving a historic leap forward for Chinese radio astronomy."
FAST's advantage is single-dish sensitivity; DESI's is sample size and precise redshifts. Combining the two systematically broke, for the first time, the old trade-off between observing deep but not wide, or wide but not deep.
References
1. NAOC / Shanghai Astronomical Observatory / SJTU et al., *The evolution of neutral hydrogen over the past 4.5 Gyr*, Nature Astronomy, published online 2026-09-01 2. Guangming Daily, "China's Sky Eye Redraws the Picture of Cosmic Evolution," 2026-09-02 (p. 8) 3. CCTV Finance / The Paper: "A Major Discovery from China's FAST," 2026-09-01 4. Xinhua: "Ten Years of Watching the Stars: FAST Redraws Cosmic Evolution," 2026-09-01 5. FAST: Five-hundred-meter Aperture Spherical Telescope, Pingtang, Guizhou, completed 2016-09-25 6. DESI: Dark Energy Spectroscopic Instrument, 4 m primary mirror, 5-year survey 7. 21 cm line: hyperfine transition of neutral atomic hydrogen, 1420 MHz (λ ≈ 21 cm)