Swift's Rescuer Ran Dry Before It Could Save the Telescope

Sometimes the most revealing part of a failure is the part that never started. NASA's Neil Gehrels Swift Observatory made its final observations on Oct. 20, 2026, while orbiting about 171 miles (275 kilometers) up. Drag from the thin air at that altitude meant its telescopes couldn't hold steady, so NASA ended the mission three days later. Swift reentered the atmosphere on Nov. 4 above the North Sumatra region of Indonesia and largely burned up, closing out two decades of watching powerful cosmic explosions.

The more pointed story is the rescue that failed first. LINK, a commercial robotic spacecraft built to raise Swift into a safer orbit, reentered Earth's atmosphere in September. The machine meant to save a falling satellite burned up before it could do anything.

The Rescuer Started Spinning

The plan was straightforward on paper. LINK would approach Swift, grab it with one of its three robotic arms, and fire its thrusters to push the telescope back up to about 373 miles (600 kilometers) above Earth, where it could keep observing for years.

In late July, LINK lost two of its three reaction wheels, the heavy spinning wheels that let a spacecraft turn without burning fuel. Combined with problems in its propulsion system, the loss sent the spacecraft into an uncontrolled spin.

Without working wheels, steering falls to the thrusters, and every burn costs propellant. That's the trap.

Engineers eventually slowed the rotation by firing thrusters against the spin. NASA reported in August that the spin had dropped from about nine degrees per second to roughly one and a half.

NASA and Katalyst Space, the company that built LINK, called off the capture and boost later in August. The team still flew LINK within 7.5 to 9 miles (12 to 15 kilometers) of Swift, deployed its three robotic arms, and tested its electric thrusters. It never got close enough to grab the telescope.

Fuel Is the Rescue Budget

A servicing spacecraft launches with a fixed supply of propellant, and engineers divide it among the job: reaching the target, approaching safely, capturing it, and moving it to a new orbit. A reserve for surprises sits on top of that, but it's usually a thin layer.

Think of a tow truck that burns most of its gas just getting out of a ditch. Even if it reaches the broken-down car, it may not have enough fuel left to tow it home. That's close to what happened with LINK. Even after engineers brought the spin under control, Katalyst said the spacecraft no longer had enough propellant to capture Swift and raise its orbit safely.

There's a real tension here. A budget that covers only the planned steps leaves nothing for recovering from a fault. But every extra pound of propellant adds mass, and heavier spacecraft cost more to launch and more fuel to move. Designers are always trading recovery margin against the cost of carrying it.

Speed Buys Time but Costs Backup

Solar activity starting in 2024 swelled Earth's upper atmosphere, and the extra drag pulled Swift down faster than anyone expected. Once NASA saw the pace, a conventional development program, which typically takes years, risked arriving too late. The agency had to move fast.

NASA awarded Katalyst a $30 million contract in September 2025. The company had less than a year to design, build, test and launch LINK, which lifted off on July 3, 2026.

Compressed schedules carry their own costs. They leave less room for repeated testing, spare hardware and planning recovery options. Public reports don't say whether that pressure contributed to LINK's faults, and I'd be careful about assuming it did. But the timeline makes the trade-off hard to ignore.

NASA has said the effort still paid off in experience. The agency learned how to build, launch and operate a servicing spacecraft quickly. Engineers also uploaded new control software to steady a damaged spacecraft using the hardware it had left, and LINK tested its arms and fired all three thrusters at once in orbit.

What Comes Next

Swift was the most pressing case, but not the only one. NASA's Hubble Space Telescope could eventually need a similar boost, and news reports suggested it would have been next in line had this mission worked.

Two complementary approaches are in development. Astroscale's ELSA-M targets satellites fitted with a docking plate before launch, which makes them far easier to grab and boost if trouble arrives later. The European Space Agency's ClearSpace-1, planned for 2029, aims to capture Proba-1, a satellite with no docking plate or capture aid, and remove it from orbit.

LINK's mission showed that a rescue takes more than a robotic arm and enough thrust to reach a target. The servicer also has to stay stable, recover from its own faults, and hold enough propellant in reserve to finish the job. Swift's rescuer fell short before the rescue could begin, and that's the lesson the industry will be watching closely as the next servicers head up.