No Robot Has Captured an Uncontrolled Satellite. The Pentagon Is Buying It Anyway.

By Nick David, Editorial Lead, SatNews

The Bottom Line:

• DIU and the Space Development Agency named three winners of Deorbit-as-a-Service design awards on Aug. 13. Six days later, NASA and one of those winners stood down from the capture that winner was already flying.
• LINK failed on reaction wheels and cold gas thrusters, bus hardware several steps upstream of the gripper, which puts the sector’s binding constraint on basic spacecraft reliability.
• No regulator or program of record mandates a standardized capture interface, and two have explicitly declined to. Every deorbit contract signed today is a contract to grab something built to be ungrabbable. Watch whether SDA specifies an interface on a future tranche.

Six days.

That is the gap between the Defense Innovation Unit and the Space Development Agency naming Katalyst Space Technologies, D-Orbit and Firefly Aerospace as winners of Deorbit-as-a-Service design contracts on Aug. 13, and NASA confirming on Aug. 19 that Katalyst’s LINK spacecraft would not capture or boost the Neil Gehrels Swift Observatory.

Katalyst appears on both dates. The gap measures how far the U.S. government’s willingness to buy has run ahead of anything demonstrated in orbit.

What broke on LINK

LINK launched July 3 out of Kwajalein Atoll on the final flight of the Pegasus XL, under a $30 million SBIR Phase III award NASA issued in September 2025. Katalyst built it in nine months. By the weekend of July 25, two of LINK’s three reaction wheels were inoperable and its cold gas thruster system had lost some function. The spacecraft entered a multi-axis spin at roughly 9 degrees per second, lost communications intermittently and reset its bus. Katalyst traces the disabled wheels to a fault in the upstream bus electrical power system, not to the wheels themselves.

By Aug. 6 the team had slowed the spin to about 1.47 degrees per second using a single electric propulsion thruster, and it still wasn’t enough margin to attempt the capture. On Aug. 19 NASA cited “an ongoing commercial spacecraft attitude control issue” and dropped it. LINK will still attempt rendezvous and proximity operations with Swift, and as of early September had not begun them. It was raising its orbit to start phasing.

These are the most ordinary components on a spacecraft. LINK never got close enough to Swift for the capture mechanism to matter.

The deorbit capability nobody has demonstrated

Northrop Grumman’s SpaceLogistics has the only repeated success, and it is a different business. MEV-1 and MEV-2 docked with Intelsat 901 and Intelsat 10-02 in GEO, grappling the liquid apogee motor nozzles of two satellites that were under control and stable throughout. Cooperative client, stable attitude, GEO, a paying operator who wants the docking to happen — that is life extension, a different problem from disposal.

Everything on the disposal side is still ahead of its first proof. Shuttle crews did grapple stranded satellites in the 1980s and 1990s, Solar Max and Intelsat 603 among them, but those clients were stabilized by ground command or by hand before capture, with astronauts on site. None of it transfers to an uncrewed servicer closing on a dead spacecraft alone.

Against that record, SDA committed $52.5 million to Starfish in January for PWSA end-of-life disposal, Space Systems Command added $54.5 million in February for a dedicated GEO servicing Otter, and DIU and SDA put roughly $8.4 million combined into three DaaS design studies in August. D-Orbit puts the potential value of its own award at $24 million, a ceiling that must count the unawarded flight phase; DIU has not published per-company splits.

That $3 billion works out to about $300 million a year, across every servicing and refueling provider on the planet.

Nobody will mandate the interface

Two regulators have looked squarely at requiring a capture interface and said no.

Maj. Gen. Stephen Purdy was blunt about the one requirement that did get written down. “It absolutely has to have a refueling requirement,” he said of RG-XX in September 2025. “I mandated that.” Nothing comparable exists for capture, which leaves the sector’s central engineering problem entirely to the market.

Joel Sercel, president and CEO of TransAstra, which is teamed with D-Orbit on its DaaS bid, named the consequence in the announcement.

That is the whole procurement in one sentence. The market is being asked to build a robot that can grab anything, for satellites deliberately built with nothing to grab, and to price that as a service.

The case for what DIU is doing

The fair defense of the DaaS structure is that it is cheap and honest about its own uncertainty. Roughly $8.4 million buys three preliminary designs through PDR, targeted for the end of 2026, and Phase II is a competitive down-select to one flight demonstration. That is risk buying, and a rounding error against a program the size of PWSA. Jared Isaacman made the same argument about Swift on the day it was cancelled: NASA “should be willing to move quickly and take smart risks when the potential return is worth it.”

He is right about Swift, which cost $30 million and bought hard data on what fails during commissioning. The trouble is what happens at the down-select. DIU’s announcement of the awards says the two agencies will weigh technical maturity, mission approach, schedule and affordability. Three of those four are assessable on paper. The fourth requires flight, and none of the three bidders has a flight record to assess. At PDR, technical maturity is whatever the design documents say it is.

What to watch

Two things.

First, whether SDA specifies a standard capture or disposal interface on a future tranche. One line in a solicitation would move the cost of disposal from the servicer to the bus vendor, where it belongs. It would also make every satellite already flying without a fixture more expensive to clean up, and advantage the bus vendors that ship one as standard fitment.

Second, whether any robotic servicer captures an uncontrolled object before the DaaS down-select. Otter Pup 2 is the nearest candidate, and it has been closing on its target since spring.

Swift, for its part, restarted science observations in late August. Its X-ray and ultraviolet-optical telescopes are back on and the Burst Alert Telescope is due back this month, on NASA’s judgment that “the potential science return is worth sacrificing a few weeks or months of orbital lifetime.” Swift drops below 300 kilometers within one to two months. It is doing astronomy on the way down.

---

About the Author

A storyteller at heart, Nick David covers space policy, satellite markets, defense, and the technologies reshaping how humanity operates beyond Earth. With a background in creative direction, brand strategy, and editorial storytelling, he brings a modern lens to complex subjects and a relentless curiosity about what comes next.