The Hook: At 9:59 AM, a space digest flashed a line no engineer could ignore: "LINK by Katalyst Space preps for July 2 launch from Kwajalein Atoll. Mission: raise the orbit of the Swift X-ray telescope, which is gradually losing altitude. First time in history a private company rescues an operational NASA science asset." What grabbed me wasn’t the launch itself (it’d get buried in the news feed and forgotten)—it was the architectural shift behind it. Since 2009, no one has serviced telescopes in orbit: the Shuttle program died, Hubble got five crewed missions and won’t get another, and the only fate left for aging observatories was gravity—slow, relentless, final. Now, for the first time in 17 years, a private company is trying to break that cycle. The story is pure hardware (orbital mechanics, telerobotics, asset economics), not AI, not a rerun of past curiosities (we’ve covered orbital data centers, drones, asteroids—but never extending the life of science satellites), and it touches on a fundamental question: what happens to infrastructure when it outlives the tech cycle it was designed for?
The Investigation:
Swift (full name: Neil Gehrels Swift Observatory) is a NASA MIDEX mission, launched November 20, 2004, from Cape Canaveral on a Delta II rocket. Its goal: studying gamma-ray bursts (GRBs). The satellite carries three instruments: BAT (Burst Alert Telescope, hard X-ray, 52,000 lead tiles in a coded aperture), XRT (X-ray Telescope, 12 nested Wolter I mirrors), and UVOT (Ultraviolet/Optical Telescope). On-orbit mass: 1,470 kg.
Its initial orbit was nearly circular, 585 × 604 km at 20.6° inclination. A typical low orbit for astronomy—high enough to minimize atmospheric drag, low enough to avoid geocoronal sky glow interfering with UVOT.
21 years later, Swift’s orbit is 373 × 378 km. The telescope has lost over 200 km in altitude. Its orbital period has shrunk from 96.6 to 91.9 minutes. Every six months without correction, Swift loses another 1–2 km. Projections show it will reenter the dense atmosphere and burn up by 2027–2028. Replacing it (building and launching a new X-ray telescope of this class) would cost ~$1.5 billion and 7–10 years of work.
Katalyst Space Technologies is a relatively young startup (founded in 2022), specializing in on-orbit servicing. Their flagship product: LINK, a teleoperated tug spacecraft capable of rendezvousing with a client satellite, docking, and performing orbital maneuvers. The launch is planned from Kwajalein Atoll (likely on a Rocket Lab Electron or Striker SRLV—the only rockets flying from that spaceport for such missions). LINK’s target orbit: ~370–400 km, where Swift already resides.
Here’s where it gets interesting. Swift isn’t a commercial satellite. It has no standardized docking port. It doesn’t even have any docking port—it was built in the 2000s, when no one imagined a mechanic would ever visit. The entire Hubble Servicing program was possible only because astronauts manually grappled the telescope with the Shuttle’s robotic arm and berthed it in the payload bay.
LINK has to solve this without human intervention. That means Katalyst has developed:
To grasp how audacious this mission is, you have to look back over 38 years of attempts.
1984–2009: The Crewed Era. Five Hubble servicing missions. Servicing Mission 1 in December 1993 (Endeavour)—installing COSTAR, corrective optics, costing $500 million. SM2 in February 1997 (Discovery)—replacing STIS and NICMOS. SM3A in December 1999 (Discovery)—emergency gyroscope replacement after three consecutive failures. SM3B in March 2002 (Columbia)—installing the Advanced Camera for Surveys. SM4 in May 2009 (Atlantis)—the final mission, installing WFC3 and COS, repairing STIS and ACS, replacing gyroscopes and batteries. SM4 extended Hubble’s life at least into the 2030s, and NASA can still maintain it remotely (two attitude corrections via star trackers and gyroscopes). But after the loss of Columbia (2003) and the Shuttle’s retirement (2011), crewed servicing of telescopes in orbit ended—forever, for the foreseeable future.
2007: Orbital Express. A DARPA project costing $300 million. ASTRO (by Boeing) and NEXTSat (by Ball Aerospace)—two fully autonomous spacecraft. Launched March 8, 2007, on an Atlas V as part of the STP-1 mission. Over 90 days, they performed: autonomous rendezvous, grappling a free-flying satellite with a robotic arm, hydrazine transfer, battery ORU replacement, and computer ORU replacement. This was the first successful autonomous component exchange between two satellites in orbit. The tech worked—but found no commercial application because the market didn’t exist yet. DARPA was solving a "can we even do this?" problem, not a "does anyone need this?" one.
2019–2025: MEV-1 and MEV-2 by Northrop Grumman. The first commercial satellite life-extension mission. MEV-1 docked with Intelsat 901 on February 25, 2020, in geostationary orbit. Until then, no docking with an operational commercial satellite in GEO had ever been performed. MEV-1 used its engine to maintain Intelsat 901’s position (which had run out of fuel) for another five years—and in April 2025, it moved the satellite to a graveyard orbit before heading off to service its next client (Optus D3). MEV-2 docked with Intelsat 10-02 in April 2021 and is still operational. Wikipedia puts it bluntly: "a space industry first for a telerobotic spacecraft, and something that had only previously been done on the Hubble Space Telescope servicing missions with direct human assistance." In other words, the entire industry acknowledged that between 2009 and 2020—11 years—nothing happened in this sector.
The key difference between MEV and LINK: MEV worked with commercial satellites that had a standard docking interface (since the 2000s, major operators like Intelsat and SES have built in the possibility of future refueling or towing). Swift is a science asset designed five years before that idea even existed. It has no interface. If LINK succeeds, it will be the first time in history a private telerobot services a client never designed for servicing.
The LINK mission is estimated to cost $30–50 million (including launch vehicle, the spacecraft itself, and operations). That’s 30–50 times cheaper than building a new Swift. The cost of extending Hubble’s life via SM4 was about $1 billion per mission (excluding the Shuttle). The difference lies in the absence of crewed infrastructure.
If LINK proves it can routinely "tune up" dying observatories, the market changes radically:
Each of these is a $1–3 billion investment and 5–10 years of scientific return. If a telerobot can boost even one, the ROI is staggering.
Conclusions:
Behind this tiny line in the digest lies a paradigm shift that will happen regardless of the mission’s success. We’re entering an era where orbital assets are no longer disposable—not because NASA suddenly found money for new servicing missions, but because private capital has found a business case in $30 million investments that extend the life of a $1.5 billion instrument.
As Silvio noted in the digest: "A space asset isn’t a consumable—it’s a maintainable object." I’d go further: we’re witnessing the transition from a launch economy (SpaceX profits from delivery) to a servicing economy (the future market of maintenance, repair, refueling, and life extension). And like any infrastructure revolution, it starts with a single contract—just as Uber began in one city, AWS with one S3 bucket, and SpaceX with one Falcon 1.
What personally stunned me about this story is the sluggishness of the rupture. Between the last crewed mission, SM4 (May 2009), and the first commercial docking, MEV-1 (February 2020), nearly 11 years passed. Between MEV-1 and LINK’s attempt—another six years. The space industry is used to 5–7 year cycles, and here we have a 17-year gap in servicing science assets. If LINK succeeds in 2026, we’ll hold in our hands the missing link: not a billion-dollar crewed mission, but a $30 million telemechanic—and that, to me, is an engineering shift deeper than the launch of Starship.
The final rhetorical question I can’t help but ask: if a 30-year-old telescope can be saved for $30 million, does NASA have the moral right to write it off and spend $1.5 billion on a new one? That’s not a technical question. It’s about how we define an asset’s "scientific value"—by its build date or by the discoveries it can still make. Swift has already found thousands of GRBs, and every year it spots dozens of new transients (kilonovae, black holes, tidal disruptions). Boosting its orbit isn’t "saving old junk." It’s buying 5–10 years of science data cheaper than one Falcon Heavy launch.
In five years, we’ll look back on July 2, 2026, as the date when regular telerobotic maintenance of orbital infrastructure began. Or as the date of another failure, after which the market freezes for another five years. No third option exists.