The robot NASA hired to lift a orbital telescope tumbled out of control

NASA bet $12 million and a storied observatory on a startup that moves fast and breaks things. The bet is currently tumbling end over end in low Earth orbit, its antenna flashing in and out of view as flight controllers fire backup thrusters to arrest a spin they never planned for.

The LINK spacecraft, built by Katalyst Space, launched July 3 on a Northrop Grumman Pegasus rocket. Its job: rendezvous with the Neil Gehrels Swift Observatory, a 2004-vintage gamma-ray hunter that has been drifting toward a fiery reentry, and shove it into a higher orbit. Swift still works. Its instruments still catch the universe's most violent flashes. But its altitude doesn't. NASA hired Katalyst to fix that — the first time the agency has contracted a private company to boost one of its own telescopes.

Three weeks into commissioning, two of three reaction wheels have failed. One thruster system is misbehaving. The spacecraft spins. Communications sputter. Katalyst principal investigator Kieran Wilson told reporters before launch that the vehicle had been "built incredibly quickly because of the urgent need to raise Swift in the next few months." He also said: "All this is challenging and risky. There're a lot of spacecraft that have had far longer development cycles with far more funding behind them that have failed for mundane reasons."

He wasn't wrong about the risk. He was wrong about the timeline. The mundane reasons arrived early.

This is what happens when you compress a decade of spacecraft development into a funding cycle. Katalyst raised its $12 million Series A in June — backed by Fortitude Ventures and Geodesic Capital — and launched a month later. The investors knew the schedule. NASA knew the schedule. The military knew the schedule. The U.S. Space Force has funded Katalyst separately, interested in a "dynamic vehicle that can service satellites or surveil rival spacecraft." That sentence should carry more weight than it does in the press release. A robot that can nudge a telescope up can nudge an adversary's satellite down. The dual-use logic is baked into the business model.

But dual-use hardware still has to work. Reaction wheels are not exotic technology. They are momentum-exchange devices flown on thousands of missions since the 1960s. Losing two of three during commissioning suggests either a batch defect, a control-software fault, or a structural vibration the rushed test program never uncovered. The thruster issue compounds it. Katalyst says it has begun corrective burns on a redundant thruster set and sees "the intended effect." NASA says it will decide whether the mission continues once the spacecraft is stabilized. That decision will come down to propellant margins. Every kilogram burned arresting the spin is a kilogram not available for the rendezvous, the capture, the boost, and the disposal maneuver.

Swift has waited 20 years. It can wait a little longer — but not indefinitely. Atmospheric drag at its current altitude is measurable. Each orbit loses a few meters. The window to act is not abstract; it is defined by physics and fuel. If LINK recovers, it still must demonstrate autonomous rendezvous, capture, and a sustained burn with a 1,500-kilogram passenger it has never touched. That sequence has never been attempted commercially. Northrop Grumman's Mission Extension Vehicle demonstrated similar logic on Intelsat 901 in 2020, but that was a geostationary servicer built on a decades-old bus with Deep Space Network tracking. LINK is a new bus, new software, new company, operating in LEO with sporadic ground contact.

The spinning antenna is a metaphor. Katalyst's communications strategy has been similarly intermittent. The company's statement to TechCrunch — "this remains an active mission, and we continue to move forward with plans to rendezvous with Swift" — is the sort of language that preserves option value for investors while the engineering team fights physics. Wilson's pre-launch candor about risk was refreshing. His company's post-failure cadence is familiar: assure, minimize, reframe.

NASA's Commercial Services office will frame this as learning. The agency has moved decisively toward fixed-price, milestone-based contracts that shift development risk to contractors. That model works when the contractor has flown hardware before. It frays when the contractor is flying its first vehicle on a mission with no margin. The Swift boost was always a pathfinder. Pathfinders sometimes find the path by falling off it.

The military interest complicates the failure analysis. If the Space Force views LINK as a prototype for inspector-interceptor missions, then the reaction-wheel anomaly is not just a mission risk — it is a capability signal. Adversaries will note that a vehicle funded for operational flexibility could not hold attitude through commissioning. Allies will note that the redundancy architecture allowed recovery burns. Both conclusions are useful. Neither saves Swift.

Katalyst's next moves are mechanical. Slow the spin. Re-establish high-bandwidth telemetry. Verify the remaining reaction wheel and the backup thrusters can hold a stable attitude for the weeks-long approach. Then prove the capture mechanism — a clamping ring designed for Swift's launch adapter — can mate without relative velocity damaging either spacecraft. Then execute a multi-burn raising maneuver that must be precise enough to leave Swift in a stable orbit for years, not months.

Each step consumes propellant. Each step assumes no further hardware surprises. The company built fast because the window was closing. The window is now tighter.

There is a version of this story where the backup thrusters arrest the spin, the remaining wheel holds, the rendezvous succeeds, and Swift watches gamma-ray bursts for another decade. That version makes Katalyst a case study in lean space ops. It makes NASA's commercial office look prescient. It makes the military's dual-use bet look wise.

There is another version where the spin returns, or the capture fails, or the boost undershoots. That version ends with Swift burning up and Katalyst writing a lessons-learned report for a contract that paid for success, not education.

The spacecraft is spinning right now. The engineers are burning. The clock is ticking. Editorials don't change orbits. But they should record who accepted the compressed schedule, who funded the dual-use logic, and who will explain the failure if the backup systems don't hold. The telescope did not choose this risk. NASA did. Katalyst did. The investors did. The military did. Swift just waits — still working, still watching, still falling.