Until several of its pointing gyroscopes failed, Kepler stared at a fixed field of the sky containing 150,000 stars.

Carter Roberts/NASA

Kepler, NASA’s planet-hunting space telescope, is dead

*Update, 30 October, 4:55 p.m.: NASA announced today that after 9 years in space, the Kepler planet hunter has run out of fuel and will be switched off in its current safe orbit far from Earth. It found a menagerie of many planet types, some unknown in our solar system and thought to be impossible. It was “the little spacecraft that could,” project scientist Jessie Dotson told a press briefing. “It did everything we asked of it, and more.” Here is our early obituary, published on 17 October:

Kepler, the NASA space telescope that almost single-handedly transformed the study of exoplanets, may have sent its last data home. Managers halted the craft's latest observing campaign last month, when its pointing became unstable—a sign of low fuel supplies. To conserve fuel, they put it in "nap mode" until a scheduled high-speed download opportunity began on 10 October. Observations are still on hold as the mission team sifts the spacecraft's vital signs for hints of the remaining fuel supply, but the end could come suddenly and without warning, says project scientist Jessie Dotson of NASA's Ames Research Center at Moffett Federal Airfield in California. "The only way to tell is to watch the spacecraft," she says. "We'll know when we know."

When the end comes, it will be a bittersweet conclusion to a prolific mission that upended theories. Before Kepler, astronomers had found only a few hundred exoplanets, mostly gas giants uncomfortably close to their parent stars. Kepler found thousands, of many sizes—including 30 tantalizingly similar to Earth. Kepler, which pioneered a novel approach to planet finding, was a gamble, says Bruce Macintosh of Stanford University in Palo Alto, California. But, "The universe cooperated, beyond anyone's wildest dreams."

Kepler's success is a vindication for William Borucki, the NASA space scientist who struggled to persuade the agency to build it. Early exoplanet discoveries came from observations of stars wobbling from the tug of large planets. Borucki suggested planets would also reveal themselves when they passed in front their host star, blocking a tiny fraction of its light. He just needed a photometer 1000 times more sensitive than any available. "You can do an awful lot with dedicated people and technology if you persist," says Borucki, now retired from NASA Ames.

Persist he did. Borucki's proposals were rejected four times before NASA finally approved the mission in 2001. "He really had a long fight," says Didier Queloz, an astronomer at the University of Cambridge in the United Kingdom who in 1995 helped discover the first exoplanet around a normal star, using the wobble technique.

The $692 million Kepler finally launched in 2009, with a 1.4-meter mirror and a 95-megapixel camera, the largest in space at the time. It stared at the same patch of sky, about the size of a fist at arm's length, simultaneously monitoring the brightness of 150,000 stars. To date, it has captured about 2650 confirmed exoplanets, and nearly 3000 more await confirmation. Among the haul are 30 rocky planets, all less than twice the size of Earth, orbiting in habitable zones where liquid water could exist. Assuming Kepler's star field is typical—and allowing for the fact that it detects only the small fraction of planets that transit—astronomers estimate the Milky Way harbors at least as many planets as stars: 100 billion.

The diversity of planets was a surprise, however. Multiplanet systems proved to be common, but they often looked wildly unlike our solar system in the sizes and orbits of their planets. "Theorists haven't caught up yet," Macintosh says. For instance, the most common planet type in the sample is a variety unknown in our solar system, halfway between Earth and Neptune in size.

The planets came so fast and furiously that, a few years into the mission, the Kepler team ended a data policy that gave mission scientists exclusive access. It began to release processed data after a few months and, last year, it began to release raw data immediately. "It made it easy for people to enter the field," Macintosh says.

There were hiccups, however. Operators found that both the camera's output and the stars themselves were more variable than predicted, making it harder to spot the tiny light dips of true Earth twins. By 2013, two of the four gyros used for pointing had failed, and the drifting spacecraft had to abandon its fixed view of the stars. Mission controllers then took advantage of sunlight pressure to partially steady the craft, but the rate of discovery slowed. As a result, the mission had to abandon a primary goal: assessing the frequency of Earth twins around sunlike stars. But given the wealth of data that hardly seems to matter. "People don't really care," Queloz says.

Successors are ready to pick up the torch. The Transiting Exoplanet Survey Satellite, launched in April to look for nearby exoplanets, has already found two, with one more awaiting confirmation. And the European Space Agency has two missions in the works: the Characterising Exoplanets Satellite, due for launch in 2019, and Planetary Transits and Oscillations of Stars in 2026. "It's an OK time to say goodbye," Macintosh says.