Hubble Spots a ‘Failed Galaxy’: Dark-Matter Cloud Cloud‑9 Has Gas but No Stars

A galaxy that exists on paper—but not in starlight

Fourteen million light-years from Earth, in a quiet patch of sky near the spiral galaxy M94, astronomers have confirmed the existence of something that looks like a galaxy in theory but not in deep optical images.

It has a massive halo of invisible dark matter. It has a cold ball of hydrogen gas thousands of light-years across. But images from NASA’s Hubble Space Telescope show it lacks the ingredient that usually defines a galaxy: stars.

“This is a tale of a failed galaxy,” said Alejandro Benítez‑Llambay, an astrophysicist at the University of Milano‑Bicocca who led the theoretical work behind the discovery. “Seeing no stars is what proves the theory right.”

The object, nicknamed Cloud‑9, is being described as the first confirmed example of a reionization‑limited H I cloud (RELHIC): a dark‑matter‑dominated, gas‑rich but essentially starless system long predicted by cosmological simulations.

The findings were published in Astrophysical Journal Letters in November 2025 and highlighted Jan. 5 at the 247th meeting of the American Astronomical Society in Phoenix during a press conference devoted to new Hubble results.

What Cloud‑9 is—and what it lacks

Cloud‑9 lies about 4.4 million parsecs (roughly 14 million light-years) away in the constellation Canes Venatici, at nearly the same distance and recession velocity as M94 (NGC 4736).

Radio data indicate it contains about a million solar masses of neutral hydrogen gas, spread across roughly 4,600 to 4,900 light-years—comparable to small dwarf galaxies. But the gas alone can’t account for the object’s structure and motions. Modeling suggests it sits inside a far more massive dark‑matter halo of about 5 billion solar masses, close to the theoretical threshold where galaxies are expected to begin forming stars.

Hubble’s cameras, however, found no sign Cloud‑9 ever lit up.

“Before we used Hubble, you could argue that this is a faint dwarf galaxy that we could not see with ground‑based telescopes,” said lead author Gagandeep S. Anand of the Space Telescope Science Institute in Baltimore. “But with Hubble’s Advanced Camera for Surveys, we’re able to nail down that there’s nothing there.”

The team reports that if Cloud‑9 hosts any stars at all, their combined mass must be below roughly 3,000 to 10,000 solar masses—orders of magnitude lower than a typical dwarf galaxy. No concentrated clump of stars or recognizable stellar population appears in the Hubble images.

From a radio detection to a decisive non-detection

Cloud‑9’s story began with radio surveys rather than optical astronomy.

In 2023, China’s Five‑hundred‑meter Aperture Spherical Telescope (FAST) detected a compact source of 21‑centimeter emission while surveying neutral hydrogen. The signal, labeled M94‑Cloud‑9, appeared as a small, isolated spot of gas near M94.

Follow‑up observations with the Green Bank Telescope and the Karl G. Jansky Very Large Array confirmed the detection and suggested the gas was relatively cold and compact, with a narrow spread in velocities—signs it was gravitationally bound rather than a loose streamer of debris.

Ground‑based optical surveys, including imaging from the Dark Energy Spectroscopic Instrument (DESI), looked for a matching dwarf galaxy but found none. Hubble then provided the critical test.

Under program 17712, Hubble’s Advanced Camera for Surveys observed Cloud‑9 for eight orbits, enabling deep color imaging and color‑magnitude diagrams—standard tools for identifying even sparse stellar populations at known distances. The result was stark: no dwarf galaxy.

“Cloud‑9 gives us a rare look at a dark‑matter‑dominated cloud,” said Andrew Fox of the Association of Universities for Research in Astronomy at the Space Telescope Science Institute, who worked on the project for the European Space Agency. “This cloud is a window into the dark universe.”

Why cosmologists expected objects like this

The RELHIC idea traces back to simulations by Benítez‑Llambay and Julio Navarro of the University of Victoria, who modeled what happens to small dark‑matter halos during and after the epoch of reionization—when the first galaxies flooded the universe with ultraviolet radiation.

That radiation heated and ionized hydrogen gas between galaxies. In low‑mass halos, gravity can be too weak to retain or cool that gas, suppressing star formation. Simulations predict that some halos might keep only a small, self‑shielded core of neutral hydrogen: compact, gas‑rich and nearly starless.

This scenario is also tied to the long‑running “missing satellites” problem. The standard ΛCDM model predicts far more small dark‑matter halos than the number of observed dwarf galaxies. If reionization keeps many halos dark, the mismatch narrows.

Anand and colleagues argue Cloud‑9 fits the RELHIC pattern: it is relatively isolated, its hydrogen appears compact and roughly spherical, and—most importantly—Hubble sees no detectable stars.

A new way to test dark matter on small scales

The discovery could offer a cleaner laboratory for studying dark matter. In most galaxies, stellar winds, radiation and supernovae can reshape gas and redistribute matter. In a system with essentially no stars, the hydrogen may trace the underlying dark‑matter halo more directly.

By modeling how the gas sits in hydrostatic equilibrium within its dark‑matter “cocoon,” astronomers can test how dense and concentrated small halos should be—details that can depend on the nature of dark matter and the physics of galaxy formation.

Cloud‑9 also anchors a key threshold: its estimated halo mass sits near the dividing line where simulations predict halos transition from starless to star‑forming. Pinning down that “turn‑on” mass helps determine how many truly dark halos might populate the nearby universe.

Some uncertainties remain. Later radio measurements suggest a broader spread in gas velocities than early data implied, potentially complicating idealized models of a perfectly quiescent RELHIC. And Hubble’s limits are not absolute: an ultra‑faint sprinkling of ancient stars could still lurk below detection.

For now, though, the team says Cloud‑9 is effectively starless.

A quiet universe that may be more crowded than it looks

Cloud‑9’s path from an anonymous radio signal to a headline Hubble result underscores how modern astronomy works across wavelengths and continents. FAST first flagged the hydrogen cloud; U.S. radio facilities confirmed it; Hubble—operated for NASA and ESA by the Space Telescope Science Institute—delivered the decisive view.

At the AAS meeting, Space Telescope Science Institute astronomer Rachael Beaton described such objects in NASA briefing materials as “houses the universe never moved into”: structures with foundations but no lights on inside.

Simulations suggest Cloud‑9 is unlikely to be unique. If ΛCDM is correct and reionization left many small halos in limbo, similar dark‑matter clouds may be scattered through the nearby universe—each holding a modest reservoir of gas but never becoming a visible galaxy.

Future radio surveys, including those expected with the planned Square Kilometre Array, could uncover more compact, isolated hydrogen clouds. Deeper follow‑up imaging with Hubble, the James Webb Space Telescope and next‑generation ground‑based observatories may help distinguish true failed galaxies from ordinary dwarfs or tidal debris.

For now, Cloud‑9 stands as rare, concrete evidence of something cosmologists have long argued must exist: a galaxy that never quite was.