A machine-learning search through NASA’s TESS observations has uncovered 10,091 candidate worlds, many around stars that were previously too faint to be examined in detail.
Scientists have found more than 10,000 possible new exoplanets hidden in NASA data, a discovery that could eventually more than double the number of known worlds beyond our solar system.
In a new study, researchers used machine learning to search through observations from NASA’s Transiting Exoplanet Survey Satellite, known as TESS. The team identified exactly 10,091 candidate planets that had not been seen before.
These objects are still considered candidates, not confirmed planets. In astronomy, possible exoplanets must be verified with additional evidence before scientists can officially add them to the catalog of confirmed worlds. Some candidates may turn out to be other cosmic objects, data errors or background noise.
Even so, the scale of the discovery is striking. Humanity has confirmed more than 6,200 exoplanets so far, according to NASA’s Exoplanet Archive. If even a large fraction of the new candidates are confirmed, the find could significantly expand the known population of planets outside our solar system.
TESS has been searching for exoplanets since 2018 by watching stars for tiny dips in brightness. These dips can occur when a planet passes in front of its host star from the telescope’s point of view, briefly blocking a small amount of light. This event is known as a transit.
Until now, TESS searches have generally focused on brighter stars, where transits are easier to detect. The new study pushed deeper into the data, examining stars roughly 16 times fainter than those typically targeted by the mission.
Using machine learning, the researchers surveyed more than 83 million stars observed during TESS’s first year of operations. Among them, they found thousands of signals that looked like possible planet transits.
The team has already confirmed one of the candidates: TIC 183374187 b. The planet appears to be a hot Jupiter, a large gas giant with a mass similar to Jupiter’s that orbits extremely close to its star, making it intensely hot.
Researchers now plan to continue testing the remaining candidates and expand the search further. The first study used only the first year of TESS data, but the team hopes to analyze the mission’s second year of observations as well.
The discovery highlights how artificial intelligence and machine-learning tools are changing exoplanet research. Instead of only relying on traditional searches, scientists can now scan enormous data sets for subtle patterns that might otherwise be missed.
The boom in exoplanet science has unfolded rapidly since 1995, when 51 Pegasi b became the first confirmed planet found orbiting a Sun-like star. Since then, missions such as Kepler and TESS have transformed the field, revealing that planets are common across the galaxy.
Future telescopes could push the search even further. NASA’s Nancy Grace Roman Space Telescope is expected to help scientists study distant worlds in greater detail, including through instruments designed to directly observe planets beyond our solar system. Longer-term plans also include the Habitable Worlds Observatory, a future mission aimed at searching for potentially Earth-like planets around other stars.
