James Webb Peered Too Close to Home… and They Wished It Hadn't

In August 2024, the James Webb Space Telescope turned away from the ancient universe and pointed at the closest sun-like star to Earth. It found a planet. Then the planet disappeared. The story doesn't begin in 2024. It begins with a star that is by almost every measure the worst possible target for the observation anyone was proposing. Alpha Centauri A sits 4.37 light years from Earth, the closest sun-like star we know of. Nearly 6000 confirmed exoplanets have been catalogued around stars across the galaxy. Astronomers had never confirmed a single one around Alpha Centauri A. Generations had pointed their best telescopes at it and found… nothing. And there were good reasons for that. The star has three properties that make it almost uniquely hostile as an imaging target. Too bright. Bright enough to overwhelm and blind sensitive detectors during acquisition. Too fast, moving across the sky at a rate standard tracking systems weren't built for. Too crowded, sitting in the densest region of the galactic plane, surrounded by thousands of contaminating stars. Three separate problems, each severe on its own. And there was a fourth. Alpha Centauri A has a binary companion, Alpha Centauri B. The two orbit each other on an 80-year ellipse, ranging from 11 to 36 times the Earth-Sun distance. That gravitational presence truncates the disk of material around Alpha Centauri A so severely that the region, save for planet formation, extends only about 0.5 astronomical units from the star. Most models suggested a gas giant in the inner system here was essentially impossible to build. Which is why, in 2017, an astronomer named Charles Beichman sat down and wrote an email to a group of colleagues, proposing exactly that. He seemed to understand how it would land. The email opened with this line. When you stop laughing, let's think about doing this project. The telescope you wanted to use hadn't launched yet and wouldn't for another four years. His argument was proximity. Alpha Centauri just lets us cheat because it's closer than everybody else. That word, cheat, is the thing to hold on to. Closeness is what makes direct imaging possible where it would be impossible everywhere else. So naturally, they pointed the most expensive telescope in human history straight at it. Webb was not the first telescope pointed at this problem. In 2019, five years before the S-1 detection, a team already found something, and then lost it. A team led by astronomer Kevin Wagner used ESO's Very Large Telescope in Chile. They were part of a project called NEAR, New Earths in the Alpha Centauri region, a collaboration between the European Southern Observatory and Breakthrough Watch. The team accumulated more than 100 hours of observations using a custom thermal coronagraph. More time staring at Alpha Centauri A than any prior campaign had ever managed, and they found a faint signal. Designated C1, it appeared at roughly 1.1 astronomical units from the star. That's almost exactly Earth's distance from the sun, right at the edge of the habitable zone. The result was published in Nature Communications in 2021, 1, classified as a candidate, and then went quiet. Never confirmed, never fully explained, filed away. The problem was the atmosphere. At the exact wavelength where a cool gas giant bleeds detectable heat, Earth's own sky glows at the same frequency. Ground telescopes are blind there. Following up on C1 required a space telescope cold enough that its own structure wouldn't outshine the signal. And in 2021, nobody had one. Wagner knew what he was sitting on. We have two detections of a point of light on opposite sides of the star that are consistent with an orbit, and are consistent with being the same brightness. Two detections. Opposite sides, five years apart. Whatever produced that light was moving, and moving in a predictable arc. The candidate sat on file for three years. Unconfirmed. Unexplained. Waiting for a telescope that didn't exist yet to be pointed at a star it was never designed to see. In August 2024, that telescope was Webb, and what it found was not the confirmation anyone expected. Aniket Sangia, a Caltech PhD student and co-author on the discovery papers, described the moment. As a direct imager, you're always confounded by artifacts you're very skeptical of anything you see but this one just popped out so clearly that moment took months of engineering to make possible Webb's mid-infrared instrument MIRI deployed a coronagraphic mask directly over Alpha Centauri a precisely engineered device that physically blocks a star's overwhelming light the way your thumb blocks the sun no standard playbook existed for a target like this one the team built the observing sequence from scratch. The first problem was just getting the star onto the mask at all. Alpha Centauri A is bright enough to saturate the detector during initial targeting, so the team executed what's called a blind offset, aiming slightly off target, then correcting, so the star never hit the sensor at full intensity during setup. A workaround born entirely of necessity. The second problem was Alpha Centauri B, not a primary target, but bright enough to flood Webb's entire field of view with contaminating light even after A was blocked. The solution? Enlist a completely unrelated star as a decoy reference, building a precise model of how stellar light flows through JWST's optics, then using that model to subtract B's contamination from every image of A. After stripping both stars' light away, layer by layer, something remained. The team called it S1, more than 10 000 times fainter than Alpha Centauri A itself, sitting at roughly two astronomical units from the star, about Saturn's distance from our own sun. And then, immediately, they tried to prove it was wrong. The team ran round after round of tests, designed specifically to eliminate S1. Could it be a background galaxy photobombing the frame? A passing asteroid? A detector artifact produced by the complex subtraction process? One by one, every mundane explanation failed. The rotation test was particularly conclusive. JWST rotated between exposures at slightly different orientations. Instrumental artifacts rotate with the telescope, while real celestial objects stay fixed relative to the star. Any artifact would shift position between images. S1 did not move. The results were submitted to the Astrophysical Journal Letters in two separate peer-reviewed papers. If S1 was real, it wouldn't just be another entry in an exoplanet catalog. It would be the nearest planet to its host star ever directly imaged, and the nearest imaged world to Earth in history. And then there was the connection to 2019. S1 appeared at roughly two astronomical units from Alpha Centauri A on one side of the star. Wagner's C1 had appeared at roughly 1.1 astronomical units, on the opposite side, five years earlier, from a completely different observatory on the ground. Same approximate brightness, opposite sides of the star, two different instruments, separated by five years, each pointing at the same region around the same star. As the researchers put it, with this system being so close to us, any exoplanets found would offer our best opportunity to collect data on planetary systems other than our own. Yet, these are incredibly challenging observations to make, even with the world's most powerful space telescope, because these stars are so bright, close, and move across the sky quickly. The team scheduled two more observations to confirm it. They turned Webb back toward Alpha Centauri A in February 2025, and they could not have known that what they were about to see would be far harder to explain than finding the planet had been in the first place. February 2025. Director's discretionary time on the web. Some of the most valuable telescope hours on Earth, typically reserved for the most urgent observations in astronomy. The team pointed JWST back at Alpha Centauri A, precisely where S1 had been. They found… nothing. April 2025. A third observation. Nothing again. In most corners of astronomy, that's exactly where a discovery quietly dies. Two empty images from the most powerful telescope ever built, filed as an unconfirmed candidate and forgotten within a year. But this team didn't stop. Instead, they ran computer simulations of millions of possible orbits, feeding in every data point they had. The 2019 C1 detection, the 2024 S1 detection, the February 2025 non-detection, the April 2025 non-detection. They asked which orbital paths could reconcile all four simultaneously. The paper described the situation in terms that make it clear the researchers understood how strange this looked. We are faced with the case of a disappearing planet. To investigate this mystery, we used computer models to simulate millions of potential orbits, incorporating the knowledge gained when we saw the planet as well as when we did not. We found that in half of the possible orbits simulated, the planet moved too close to the star and wouldn't have been visible to Webb in both February and April 2025. In other words, in roughly half of all possible orbital configurations, the planet is not gone. It simply moved inside the coronagraph's inner working angle, too close to the star for Webb to isolate. It went dark in both follow-up windows, not because it vanished, but because its orbit carried it back into the glare. The best fit orbit emerging from those simulations is a Saturn-mass gas giant with a period of 2 to 3 years and an eccentricity of roughly 0.4. It does not travel in a circle, but swings dramatically closer to and farther from Alpha Centauri A on every orbit. Any orbit must also remain gravitationally stable in the presence of Alpha Centauri B, pulling from somewhere between 11 and 36 astronomical units away. An unstable planet in a binary system like this would eventually be flung out entirely. The simulations that survive both constraints, observational record and the stability requirement are what remain. When all you have is a dot of light next to a star, you may or may not be looking at an actual planet, one of the researchers observed. But when you see a dot of the same brightness several years later on the opposite side of the star, the story becomes much more compelling. The disappearance, it turned out, was not evidence against the planet. It was consistent with the orbit. The two non-detections, the two detections, everything fit. The case was, by any reasonable measure, compelling. Senge noted, the observations and orbital simulations provide us with a prediction for where the planet will be in the future, and the best time for us to go back and look at it will be in fall of next year. We look forward to that opportunity, to confirm its identity beyond any doubt. There was just one problem. The history of this star has a pattern, and it does not favor the optimists. Alpha Centauri has a documented history of planetary candidates that did not survive scrutiny. In October 2012, astronomers announced a world called Alpha Centauri Bb, described at the time as the nearest exoplanet ever found, a hot rocky planet orbiting Alpha Centauri Bb on a period of just 3.2 days. It generated significant press coverage. Subsequent analysis confirmed it did not exist. It was a data artifact, a pattern that appeared in the noise and was mistaken for a signal. That history matters because S1 appears in heavily processed data that required subtracting the light of two stars using a model built from a third unrelated reference star, a chain of assumptions where any weak link could produce a false positive. The team knew this and ran round after round of elimination tests, specifically designed to prove S1 wrong. But the history of the field suggests that awareness of a risk does not make you immune to it. The statistical picture is one of genuine ambiguity. Combining the astrometry from the 2019 S1 detection and the 2024 S1 detection, the researchers calculated approximately a 52% probability that both represent the same orbiting planet. That is technically a high bar for direct imaging of short period planets around bright stars. by another measure, a coin flip. Elizabeth Matthews, an astrophysicist at the Max Planck Institute for Astronomy in Germany, who was not involved in the research, framed it carefully. It's a really exciting candidate, quite an intriguing candidate. The authors work hard to make a case for why this, frankly, small and faint blob of light is believable. But I think there are still some open questions that need to be answered to really be 100% sure. That skepticism is fair, but the team had a response to it, and it is worth taking seriously. Even if S1 turns out not to be a planet, the observation campaign itself already proved something no one had demonstrated before, that Webb could reach this depth, at this star, through this level of contamination. One team member put it directly, all data, whether it's a detection or not, tells you something. Even if this ultimately is a non-detection, I think it's a really important paper in understanding the true capabilities that JWST's MIRI can push to, and I think that's just as important. In other words, the worst case for S1 is still a discovery. The floor is not zero. But here is the part that makes the skepticism worth enduring. If S1 is real, it does not just add one more name to an exoplanet catalog. It breaks five separate records simultaneously at the nearest solar-type star we have ever found, in a place where, according to our best models of planetary formation, no planet of this kind should be able to exist. The researchers stated it plainly in the paper, a confirmation of the S1 candidate as a gas giant planet orbiting our closest solar-type star, Alpha Cen A, would present an exciting new opportunity for exoplanet research. Such an object would be the nearest, coldest, oldest, shortest period and lowest mass planet imaged in orbit around a solar-type star to date. Five records, simultaneously, at the nearest solar-type star in the sky. The temperature of this object, if it is a planet, sits around 225 Kelvin, roughly minus 48 degrees Celsius. That makes it far more analogous to Saturn or Jupiter than to the blazing hot Jupiters that typically show up in direct imaging campaigns. Those tend to be young and intensely luminous, because they haven't had time to cool. This would be the first time we'd photographed something that genuinely resembles our own solar system's gas giants from the outside. Its existence would force a rethinking of how planets form in binary systems, because the accretion safe zone around Alpha Centauri A extends only to 0.5 astronomical units, while this object sits at two astronomical units. The paper says it directly. Its very existence would challenge our understanding of the formation and subsequent dynamical evolution of planets in complex hierarchical systems. The orbit carries a sharp cost, though. An eccentricity of 0.4 means the planet swings through Alpha Centauri A's habitable zone on every circuit. As Beichman noted, giant planets are the big movers and shakers of the planetary system, so if you have an eccentric giant planet, that doesn't bode well for the survival of terrestrial planets. Any rocky worlds that might have formed in the habitable zone, were probably ejected long ago. But moons are a different calculation entirely. Wagner observed, no indications of moons just yet, although if we look at Jupiter or Saturn in our own solar system? They're both orbited by this rich system of moons, so I would be surprised if it doesn't have moons. Moons orbiting a gas giant that swings periodically through a habitable zone could, in principle, retain liquid water during the warmer parts of that orbit. It's a narrow thread of possibility, but it points at something James Cameron already put on screen. In Avatar, the habitable moon Pandora orbits a gas giant called Polyphemus around Alpha Centauri A. The real candidate's orbital position is uncomfortably close to that fiction. A probe traveling at 1%, of the speed of light would reach Alpha Centauri in 430 years. That's a long time. It's not infinite. The species that eventually makes that journey will want to know what they're going to find before they go. Which brings us back to the telescope, and to a specific date, August 2026. The orbital model that reconciles all four data points the 2019 detection the 2024 detection the February non-detection the April non-detection predicts exactly one viable next observing window Sangi and Beichman stated it directly the observations and orbital simulations provide us with a prediction for where the planet will be in the future and the best time for us to go back and look at it will be in fall of next year. We look forward to that opportunity to confirm its identity beyond any doubt. That paper was published in 2025. The window it describes is August 2026. Getting there is not automatic. Valuable web time must be requested, reviewed and approved in competition with every other program trying to access the most oversubscribed telescope ever built. But Webb is no longer the only option. NASA's Nancy Grace Roman Space Telescope carries a coronagraph specifically custom-built for imaging large exoplanets around nearby bright stars. It was expected to begin operations around the same time and could independently attempt the same observation with a fresh instrument and entirely different optics. If August 2026 confirms S1's simultaneous records the nearest directly imaged planet orbiting a solar type star and Charles Beichman's 2017 email the one that opened with when you stop laughing let's think about doing this project sent four years before the telescope launched about a star four light years away ends in vindication if August 2026 finds nothing the orbital mechanics still permit windows further out, or S1 was always dust scattered by the optical chain, or an artifact of a detector being pushed past its design specifications, producing a convincing pattern in the noise. The history of planetary candidates around Alpha Centauri has one consistent feature across every prior attempt, and it's worth naming. Every one of them has disappeared. Alpha Centauri A is the third brightest star in Earth's night sky. From the northern hemisphere, it never clears the horizon. In the southern sky, on any clear night, it is a single point of steady light, unremarkable to the eye, indistinguishable from thousands of others. It has been there every night for the entirety of human history. Every generation that has ever existed on this planet looked up at it, without knowing whether anything orbits it. We are the first generation with a telescope capable of answering that question. And the answer is due in August 2026, which, depending on when you're watching this, is either very soon, or already this side.