NASA is sending a robot to grab a dying telescope it never built to be saved

Space 4 min 80 sources

A $30 million rescue mission launches to tow the 21-year-old Swift telescope back to a safe orbit, because Swift was built without the thrusters to save itself. Plus $600M for four Moon landers, a plan for 100,000 data-center satellites, and a spacewalk to fix the station's robotic arm.

Key takeaways

NASA is spending $30 million to send a private robot to grab and tow its 21-year-old Swift telescope back to a safe orbit, because Swift was built with no way to save itself.
The same week, NASA put nearly $600 million toward four private Moon landers, and a new startup asked to launch 100,000 satellites to run AI computers in space.
A thread runs through it: space is shifting from build-and-discard toward maintain-and-repair, and we are only now building the tools to do the fixing.

A robot arm reaches for a falling telescope

Early Tuesday, a Northrop Grumman Pegasus rocket dropped from the belly of a jet over the Marshall Islands and flew to orbit for the last time [7]. Its cargo: a small robotic spacecraft called LINK, built by Arizona’s Katalyst Space Technologies, sent to catch a telescope that is slowly falling out of the sky [7].

The telescope is NASA’s Neil Gehrels Swift Observatory — a $500 million instrument launched in November 2004 to watch gamma-ray bursts, the brightest explosions in the universe [7]. More than 20 years on, Swift still does good science [7]. But its orbit has begun to sag. Recent solar activity has puffed up the upper atmosphere, and the thin gas at Swift’s altitude drags on it a little harder each pass [7]. Left alone, it burns up.

Here is the catch that makes this a story: Swift was never built to be saved. It carries no thrusters to lift its own orbit, and no grapple points for another craft to grab [7]. When engineers designed it in the early 2000s, the plan was simple — the telescope works until it dies, then re-enters. Nobody planned a rescue.

So NASA is improvising one. LINK, about 4.9 feet tall with three robotic arms, will spend two to three weeks studying Swift up close to find a safe place to grip a spacecraft that was never designed to be gripped [7]. Then it will fire gentle ion thrusters — engines that push with a faint, steady stream of charged particles — to raise the pair back toward Swift’s original altitude of about 373 miles over several months [7]. Ion thrust is weak but patient; it cannot shove, only nudge, so the climb is slow [7].

The whole mission cost NASA $30 million [7]. Replacing Swift’s abilities from scratch would cost far more [7]. NASA selected Katalyst for the job in September 2025, giving the company less than a year to design, build and test LINK [7]. If it works, LINK will be the first private spacecraft to grab an uncrewed U.S. government satellite [7].

Building the Moon’s supply chain, one lander at a time

While one team fights to keep an old telescope alive, another is spending to plant new hardware on the Moon. On Tuesday NASA awarded nearly $600 million for four robotic lunar lander missions, split among three companies, as part of its long-term plan to build a base near the lunar south pole [2].

The awards run through NASA’s Commercial Lunar Payload Services program — CLPS, the agency’s approach of hiring private landers to fly its instruments rather than building government spacecraft for each trip [9]. The idea is a delivery service to the Moon: NASA buys the ride, companies own the vehicle. Several of these landers have reached the Moon in the past two years; some have tipped over on landing, a reminder that touching down gently on another world is still hard [9].

A plan to move the data center to orbit

A five-month-old startup called Orbital asked U.S. regulators in late June for permission to launch up to 100,000 satellites — not to beam internet, but to run computers [5]. The plan is 10 gigawatts of computing power in orbit to feed the rising electricity hunger of artificial intelligence [5]. Ten gigawatts is roughly the output of ten large nuclear reactors.

Treat this as a filing, not a fleet. Asking a regulator for permission is cheap; building and launching 100,000 spacecraft is not, and no one has done anything close [5]. The pitch rests on a real pressure — data centers on Earth are straining power grids and water supplies — and on the promise of free solar power and cold vacuum in space [5]. Whether the economics hold is the open question.

Fixing the arm that fixes everything

Up on the International Space Station, astronauts spent this week preparing for and carrying out a spacewalk to repair Canadarm2, the station’s 57-foot Canadian-built robotic arm [17][41][55]. The arm catches arriving cargo ships and moves equipment and crew outside the station — the tool that maintains the outpost. When the tool that does the maintaining needs maintaining, someone has to go outside and do it by hand [55].

02 · Lesson · why it matters

The quiet assumption inside everything we build: someone else will clean it up

Swift was doomed not by a flaw but by a default — it was built to be thrown away, because in 2004 nobody imagined it could be rescued.

A telescope that could not ask for help

Swift is not broken. Its instruments still work, its science is still useful after 21 years, and it is one of the best gamma-ray watchers humans have ever flown. It is going to burn up anyway. Not because something failed — because of something that was never there. Swift has no thrusters to lift itself, and no handholds for anyone else to lift it. It cannot save itself, and until this week, it could not be saved.

That absence is the whole story. And it wasn’t an oversight. When engineers designed Swift in the early 2000s, they made a reasonable choice: don’t add the weight, cost, and complexity of self-rescue hardware, because that is not how the game works. Satellites go up, they do their job, they die, and — if they matter — someone builds a new one. The plan for the end of Swift’s life was that there would be no plan. It would fall.

The default that looked like a law

Here is where the pattern starts, and where it reaches far past space. Every satellite of that era carried the same unspoken assumption: I will not be serviced. Nobody wrote it on a whiteboard as a rule. It was just the water everyone swam in. Building to be repaired costs more today, and the payoff comes years later, to someone else’s budget. So the cheap default won, over and over, until it stopped looking like a choice and started looking like a fact of nature — the way things simply are in orbit.

That is the second face of the whole: a system has a shape. Some parts set the terms the rest live under. “Satellites are disposable” wasn’t handed down by physics. It was a decision — made by people optimizing their own launch, on their own timeline — that hardened into the terms every later mission had to accept. Swift’s designers didn’t choose to doom it. They inherited a default and passed it on, exactly as the people before them had.

Who pays for a choice nobody remembers making

Look at who is inside this now. NASA is spending $30 million and the last flight of a historic rocket to grab a telescope that was never built to be grabbed. A young company had less than a year to invent the arms that could hold a spacecraft with no handles. The rescue is slow, delicate, and improvised — because the object being rescued was engineered on the assumption that this day would never come.

The cost of the old default did not vanish. It moved. It travelled forward twenty years and landed on a different budget, a different team, a harder problem. This is what a design default does when it turns out to be wrong: it doesn’t announce itself. It waits, quietly, until the bill comes due to whoever is holding the system when it fails.

The same shape, closer to home

Now widen it, because Swift is not really about telescopes. Almost everything we build carries a quiet assumption about who maintains it and who cleans up after — and that assumption almost always favors the person building today over the person inheriting tomorrow.

The code written to ship fast, with no plan for who patches it in five years. The bridge, the pipe, the grid built to a lifespan, with the replacement left as a problem for a government not yet elected. The pension, the landfill, the warming atmosphere — each one a system where the cheap default was chosen by people who would not be the ones holding it when the orbit sagged. We are all Swift’s designers, and we are all the rescue team. We inherit defaults we never voted for, and we hand our own down without noticing.

What the rescue actually teaches

The good news in this story is real: we are learning to reach the things we abandoned. LINK is one of a small wave of servicing craft — machines built not to explore but to repair, refuel, and rescue what is already up there. Space is slowly shifting from build-and-discard toward maintain-and-mend. That is genuine progress, and it is worth naming.

But the deeper lesson is humbler than that. The rescue exists because a reasonable choice, made by careful people, turned out to carry a cost they could not see from where they sat. They weren’t careless. They were doing the sensible thing under the terms they were handed. That is the uncomfortable part: the default that dooms a system usually looks, from the inside, exactly like good engineering.

So the useful question is not “who was foolish enough to build Swift this way?” No one was. The useful question is quieter and points at us: what are we building right now to be thrown away — and who, twenty years from now, will have to invent the arms to catch it?

03 · Lab · your turn

Build It To Be Saved

Rehearse choosing the cheap default versus paying now to make a system rescuable, and feel who inherits the bill twenty years later.

04 · Hope · carry this

For decades we built things to be used up and left behind. Now, quietly, we are learning to go back for them — to catch what was falling, mend what still works, and give a twenty-year-old machine a few more years of looking outward. The reach we once spent only on leaving is starting to turn toward keeping.