The reddest dot in Webb's archive turned out to be a new kind of object

A dot too red to be anything we knew

Astronomers say they have found a new class of cosmic object. They are calling it a black hole star [1].

The object is named MoM-BH*-1. It is the reddest thing in the James Webb Space Telescope’s image archive, sitting in the constellation Cetus [1]. Its light left when the universe was about 660 million years old [1][2].

Nothing in the picture gave it away. It looked like one more faint red smudge. What settled it was the spectrum - the breakdown of its light, wavelength by wavelength.

That light is almost entirely red, then drops off abruptly below a certain wavelength [1]. Rohan Naidu, who led the team, said there is no comparison among any known class of objects [1].

Writing in Nature, the team concludes the object is a black hole swaddled in dense gas [3]. The gas is thick enough that the black hole itself cannot be seen. Energy from the feeding black hole lights the cloud from within, so the whole thing radiates like an enormous star [2].

The output does not fit a star. The Guardian reports it releases roughly 100 billion times more energy than any known star can produce [1]. That is black hole territory, not nuclear fusion.

Why it matters: Webb keeps turning up faint “little red dots” in nearly every deep image, and nobody was sure what they were [1]. This one appears to sit alone rather than buried inside a galaxy, so its light is not mixed with a galaxy’s [2]. That makes it a clean template. The team modelled what its spectrum would look like after it merges with a nearby galaxy, and got something that resembles the little red dots already catalogued [2]. Naidu suspects these are the swaddled first phase of the supermassive black holes now sitting at the centre of galaxies [1].

The same trick, five different jobs

Every element absorbs and emits light at its own fixed set of wavelengths. Read those, and you can name what the light passed through. That one trick did a lot of separate work this week.

Webb pointed its mid-infrared instrument at IRS 3, a dying star just 0.55 light-years from Sagittarius A*, the black hole at our galaxy’s centre [4]. Sagittarius A* carries the mass of about 4 million suns [4]. Combining the star’s spectrum with models of its surroundings, the team mapped shells of silicate dust reaching some 10,000 times the Earth-Sun distance [4]. Temperatures fall from about 927 degrees Celsius near the star to minus 173 at the outer edge [4]. They also found clear evidence of water - a first for this star [4].

The galactic centre is drenched in radiation, so molecules were not expected to last there. Co-author Macarena Garcia Marin said the detection shows molecular material can survive in that environment [4].

In Chile, a telescope still under construction caught gas thrown off by a dying star drifting back into the galaxy [5]. MOTHRA combines hundreds of off-the-shelf telephoto lenses into a single compound eye [5]. Fewer than 200 of a planned 1,140 lenses were installed when it was pointed at the Helix nebula during tests in November 2025 [5]. Reported in Nature this week, the observations show carbon and oxygen forged inside a star returning to the raw material for the next generation [5].

A Martian rock found in Algeria in 2019 was finally dated this month at 1.27 billion years old [6]. NWA 13441 fills a long gap in the Martian record between 600 million and 2.4 billion years ago [6]. Its neodymium isotopes resemble those in chondrites, the unmelted rocks common when the solar system formed [6]. That blend suggests deep parts of Mars have gone largely undisturbed since the planet formed [6].

Hubble has now sorted 200 million Andromeda stars into 300-light-year squares and read the colours in each one [7]. Blue means recent star formation, so the colours reconstruct the galaxy’s history square by square [7]. Star formation there is winding down, and the small companion galaxy M32 is the likeliest suspect [7].

And 115 million light-years away, astronomers spotted a thin band of stars unravelling from a globular cluster [8][9]. It is the first such stream seen outside the Milky Way [9]. Those stars follow nearly the same orbit, and that orbit is shaped by the galaxy’s gravity [8]. Model the orbit and you weigh the galaxy; subtract the visible mass and the remainder is dark matter [8].

One quieter note: the 64-metre Sardinia Radio Telescope, which made the lowest-frequency fast radio burst detection yet at 328 MHz, is offline for an upgrade until at least September [22]. It observes from 300 MHz to 116 GHz and doubles as part of ESA’s deep space network [22].

The wall: reading the air of another Earth

The tool has a hard limit, and it shows up on the one target everyone wants.

When a planet crosses in front of its star, a sliver of starlight passes through the planet’s air. The gases there subtract their own colours on the way out [10]. For a hot, puffy giant like WASP-39 b, this works well - Webb has already read its atmosphere in detail [10].

For an Earth-sized planet around a Sun-like star, the signal is roughly one part per million [10]. A very faint fingerprint on a very bright light.

The usual answer to a faint signal is to stack many measurements until the noise averages out. That fails here. An Earth-like orbit takes a year, so an Earth twin passes in front of its star once a year [10]. A decade of waiting buys ten attempts. Each one has to be good enough on its own.

Which leaves the mirror. You need something around 30 metres across, in space [10]. Webb’s mirror is 6.5 metres, took decades, and had to unfold itself on arrival [10].

Jian Ge and colleagues at Shanghai Astronomical Observatory have proposed a way around that [10]. Their concept, Life 2.0, is 900 one-metre telescopes flying as one distributed array [10]. Each carries a miniature spectrograph and its own low-noise detector; the spectra are combined afterwards [10].

The 900 is not arbitrary. Light-collecting area rises with the square of the diameter. Thirty divided by one, squared, is 900 - the number of one-metre units it takes to match a 30-metre mirror. On the same arithmetic, a 30-metre aperture gathers about 21 times the light of Webb’s 6.5 metres.

The appeal is manufacturing. One vast mirror is a bespoke, slow, terrifying build; 900 identical small ones are a production line [10]. Waveguide spectrograph prototypes have already reached 40 to 66 per cent throughput [10].

It is a concept paper from a conference, not a mission plan [10]. The authors name detector drift, instrument stability and the stars’ own variability as problems still to solve [10].

On the pads

SpaceX launched two Falcon 9 rockets 38 minutes apart on Saturday night, its tightest gap yet [11]. One carried eight Globalstar satellites from Cape Canaveral; the other flew a classified Space Force mission from Vandenberg [11].

China had a worse week. A Long March 7A exploded about 85 seconds after lifting off from Wenchang on 10 August [12]. Vehicle and payload were lost [12].

Rocket Lab said the window for a first Neutron launch this year is narrowing [13]. Chief executive Peter Beck said the rocket should reach the pad in the fourth quarter, but static-fire testing comes first [13]. The company also unveiled GHOST, a launch site packed into standard shipping containers [14].

Astra Space is attempting a comeback, seeking $250 million at a $1 billion valuation [15]. It was worth more than $2 billion in 2021, then taken private in 2024 for $11.25 million after repeated launch failures [15].

Virgin Galactic pushed the first commercial flight of its new Delta-class plane to February 2027 [16]. Chief executive Michael Colglazier said no single issue caused the slip - just small delays across hundreds of installation tasks [16].

NASA administrator Jared Isaacman said he is extremely confident Artemis 3 will fly in 2027, and stacking of the SLS rocket has begun at Kennedy Space Center [17]. After a visit to Idaho National Laboratory, NASA says it remains on track to launch a nuclear electric propulsion demonstration by late 2028 [18].

Rocket Lab also won a $397 million Space Force contract for threat-tracking satellites it calls Flatellites [19]. NASA formally invited India’s space agency to join its lunar base programme [20]. And South Korea named space one of seven national technology projects, targeting a moon landing by 2030 [21].