18 SEP 2026 — Astronomers have confirmed a Jupiter-mass planet less than a million years old, roughly five times younger than the previous record holders. It has been sitting in a telescope archive since 2018.
Elias 2-24 b orbits its star at 55 times the Earth-Sun distance, which is where the problem starts. Standard theory needs about five million years to build a planet that size at Jupiter's distance, and this one is far further out and far younger.
What was found
The planet sits in a disk of gas and dust around Elias 2-24, a young star about 450 light-years away in the Ophiuchus star-forming region. The work was published in The Astrophysical Journal Letters on 16 September, led by Andrea Bernardi, a doctoral candidate at Universidad Diego Portales in Chile.
It is still accreting. For observers that matters, because a planet actively pulling in material shows a formation stage rarely caught directly. Usually it is inferred afterwards.
The location matters as much as the age. Protoplanetary disks show concentric gaps, and the long-standing assumption is that planets carve them. "The planets should be found within the gaps, since they are carving them. And that's exactly where we found Elias 2-24 b," Bernardi said.
The discovery is in the re-analysis
New telescope time did not find this. Keck Observatory observed the system with its near-infrared camera in 2018 and again in 2020, and the 2018 data showed only a faint signal. Bernardi returned to the archive years later with better processing and improved models, and the signal resolved into something consistent with a forming planet. Data from the Very Large Telescope and ALMA supported it.
John O'Meara, chief scientist at Keck Observatory, put it plainly: "The observations existed for years, but new techniques, improved models and a fresh look at the data revealed something extraordinary."
That is an argument for funding archives. They are unglamorous next to new instruments, and they keep producing results like this one. The Keck Observatory Archive is a NASA-funded partnership with the NASA Exoplanet Science Institute at Caltech.
A candidate that was disputed
This object has a history, and the announcement is a resolution rather than a first sighting. Something at this position had been argued over before, variously proposed as a planet, an imaging artefact, or a background star happening to line up. Alice Zurlo worked on the system in 2018, and a separate 2023 paper examined kinematic and thermal signatures of what it called a protoplanet candidate around the same star.
The new work adds motion over time. Two epochs of Keck data, 2018 and 2020, let the object be tracked rather than merely seen. An artefact does not move with the disk, and a background star does not move with the star. That is what moved it from candidate to confirmed.
Why five million years is the problem
Core accretion, the leading account of how giant planets form, is a gradual process in which solids stick together into a core that, once heavy enough, captures gas. The timescale works at Jupiter's distance given roughly five million years.
Further out, formation should be slower, not faster. There is less material, orbital speeds are lower, and core-building collisions are rarer. A Jupiter-mass planet at 55 AU in under a million years is the opposite of what the model predicts on both counts at once.
Lucas Cieza of the Instituto de Estudios Astrofísicos was direct about it: the models already struggled with the previous record holders, and "Elias 2-24 b shows us that even our best planet-formation models are still missing some important processes."
The previous record was a four-way tie — two planets at PDS 70 and two at WISPIT 2 — all more than five million years old. This is not an incremental improvement on that figure.
What to watch
The next step is spectroscopy. The team is continuing to observe, and a spectrum would give the atmosphere, the temperature, a better mass and some sense of how fast it is still growing — the quantities that turn a detection into a test of formation models.
The detection sits at the edge of what current instruments can do, a useful reminder before treating the age as settled. Ages of very young stars are model-dependent, and a planet found at the limit of detectability carries more uncertainty than a clean number suggests.
The broader question is how many other archives hold the same kind of result. If re-analysis of 2018 data can produce the youngest known planet, plenty of comparable objects may already be recorded and simply have not been looked at with current techniques.
View author profile → · Editorial policy
