Katalyst Space

The $12 million figure is the company-stated June 2026 raise led by Geodesic Capital with Fortitude Ventures. Post-money valuation was not published. The Form D amount is cash sold in a 2024 Reg D offering by Katalyst Space Holdings, Inc., not a current enterprise value. Employee count is CEO Ghonhee Lee’s August 2026 description of a 60-person company; LinkedIn lists a 51–200 band and about 58 named employees. No audited revenue, margin, cash or backlog is public. Aggregator “FY2025 revenue” figures that reprint the $30 million NASA award as sales are not used here.¹⁸¹⁶

Katalyst Space Technologies is a privately held Arizona in-space servicing and space-domain-awareness company. It builds robotic spacecraft that try to rendezvous with satellites that were not designed to be docked with and sells modular hardware and software meant to be installed after launch. The public story in 2026 is the LINK vehicle: designed, built, tested and launched in about nine months under a NASA Small Business Innovation Research (SBIR) Phase III award of about $30 million, then stripped of its capture-and-boost objective after attitude-control failures. Headquarters remain in Flagstaff. The flight hardware shop is the former Atomos Space plant in Broomfield.¹²⁶

Katalyst sells the idea that a satellite does not have to die with the hardware it launched with. In plain language the company builds three layers. First, a robotic vehicle that can find another spacecraft, fly close to it and in the original plan grab it. Second, bolt-on modules that add sensing or inspection to a host that never had those ports. Third, software that classifies objects and stitches together other people’s sensors.

The national-security relevance is dynamic space operations: the ability to inspect, relocate, upgrade or dispose of an asset after it is already on orbit. The Pentagon has been writing that requirement into SpaceWERX challenges and Defense Innovation Unit (DIU) studies. NASA used the same industrial argument when it hired Katalyst to chase the decaying Neil Gehrels Swift Observatory.

Demonstrated fact and company claim are not the same thing. LINK reached orbit on 3 July 2026. It did not capture Swift. It did not boost Swift. As of late August 2026 the vehicle was still being flown for residual rendezvous and proximity operations (RPO) and imaging. That is flight activity. It is not a completed servicing mission.²³

LINK

Low Earth orbit robotic servicing vehicle built for the NASA Swift Boost mission. Company and trade descriptions put the spacecraft at roughly refrigerator size, about 425 kg launch mass, with three robotic arms, cameras and sensors, and xenon electric propulsion. Launched 3 July 2026 at 04:36 EDT on a Northrop Grumman Pegasus XL from Kwajalein Atoll. COSPAR 2026-152A. Capture and boost were cancelled on 19 August 2026 after the loss of two of three reaction wheels and the consumption of extra propellant to hold attitude. Residual RPO and imaging of Swift remain the stated work. LINK is a flown vehicle, not a catalog product with a published unit price.²⁴⁷

NEXUS

Planned multi-mission GEO-class robotic servicer. Company materials describe roughly double the power, mass and delta-v of LINK. Intended first flight second half of 2027 on Ariane 6. Stated demonstration sequence: install a SIGHT module on a U.S. Space Force satellite referred to in trade reporting as Rooster, then conduct additional government RPO and SHIELD inspection work, then commercial servicing. Also the vehicle named in Katalyst’s August 2026 DIU / Space Development Agency (SDA) Deorbit-as-a-Service (DaaS) selection. NEXUS has not flown.¹⁹¹¹

SIGHT

Modular space-domain-awareness upgrade. Company page: an independently powered and communicating sensor package that can be installed before launch or retrofitted on orbit to watch nearby resident space objects and support collision avoidance. Planned as the first hardware drop on the 2027 NEXUS GEO sequence. No independent flight-heritage certificate is published for SIGHT as a standalone article.¹²

SHIELD

Deployable scout architecture for close-range inspection and defensive space operations. A base station is described as installable on prepared or unprepared hosts; a free-flying scout would then inspect the host or a nearby object. Subject of a 2024 Air Force SBIR Phase I titled “SHIELD-1 Mission.” Not a demonstrated on-orbit capture system.¹³¹⁸

ARC

Machine-learning analytics suite for classifying resident space objects, detecting off-pattern maneuvers and cueing analysts. Originates in a 2022 Air Force SBIR Phase I on Advanced RSO Characterization. Software product. No public performance benchmark against operational Space Force catalogs is cited here.¹⁴¹⁸

ICON

Software concept for networking otherwise disconnected commercial and government space-domain-awareness sensors, including cislunar tracking algorithms and automated custody handoffs. Subject of a 2023 Air Force SBIR Phase I. A network product. Not a fielded sensor grid.¹⁵¹⁸

Quark / Atomos heritage

Reusable orbital-transfer vehicle line acquired with Atomos Space in March 2025. Atomos flew a March 2024 LEO demonstration with Quark-LITE and a target called Gluon; that mission had commissioning problems. The Broomfield plant that built Quark is now Katalyst’s hardware floor. Quark is heritage and industrial capacity, not the current public brand on the Swift mission.⁶

Near-term cash is government mission work. The NASA Phase III is the largest disclosed number on the page. Air Force and SpaceWERX SBIRs paid for the software and upgrade modules that now sit in the catalog. The June 2026 $12 million raise is earmarked by the company for NEXUS and a 2027 GEO demonstration, not for replacing the cancelled Swift boost.

Three levers matter over the next one to three years. First, whether residual LINK operations produce enough flight data to keep NASA and the Space Force writing follow-on statements of work after a public abort. Second, whether DIU and SDA take DaaS from a shared Phase I study into a funded flight. Third, whether a GEO NEXUS mission in 2027 is a real payload on Ariane 6 or a date on a briefing chart.

Commercial servicing revenue is a claim, not a disclosed book. Intelsat appears in a 2024 partnership note. No commercial contract value is public. Backlog is not published. Production capacity is the Broomfield highbay plus a Flagstaff design office and a reported Washington presence. Throughput in units per year is not disclosed.¹⁶

Katalyst forced a decision that most servicing startups never reach: fly the vehicle or keep raising against a slide deck. It flew. The calendar from September 2025 award to July 2026 launch is the part of the story that holds. The part that does not hold is the last act. Reaction wheels and propellant margin are unglamorous and they are why capture missions fail.

For national security the implication is mixed. The United States still needs a commercial way to inspect, relocate, upgrade and dispose of satellites that were never built with a docking ring. LINK put a U.S. company next to that problem on a NASA clock. The August abort is now part of the flight record that DIU and SDA will read when they decide which DaaS vendor gets a Phase II. A cancelled boost is not a cancelled company. It is a data point that the easy half of the mission is launch and the hard half is control.

For capital the implication is sequence. The $12 million GEO raise closed before the abort. The next check, if there is one, will be written against NEXUS milestones and against whatever residual RPO LINK can still show. There is no public valuation to mark to market. There is no audited book. Anyone treating the NASA award as recurring revenue is reading a contract as a franchise.

No internal Defense Briefing company-profile slug for Starfish, Astroscale, Firefly, D-Orbit or Northrop servicing units is verified in the current library index. Do not invent those routes. Adjacent live private profiles for industrial context only: Blue Origin and Maverick Space Systems.

The moat is not heritage. Northrop already docked in GEO. Astroscale has flown removal demonstrations. Atomos, now inside Katalyst, flew a 2024 test that did not go cleanly. What Katalyst can actually claim is speed under a NASA Phase III and a modular product map that tries to sell SDA upgrades and software even if the tug is late.

What is durable

Nine months from award to a Pegasus stack is unusual in U.S. civil space and is the reason investors and DIU are still in the room after the abort. The Flagstaff-plus-Broomfield split gives the company a design office and a 20,000 sq ft integration floor it did not have to build from zero. SBIR work on SIGHT, SHIELD, ARC and ICON predates Swift, which means the catalog is not a single-mission rebrand.

What is not

Capture of an unprepared government satellite remains unproven for this team. Attitude control on LINK failed in a way that consumed the boost margin. Capital is thin next to Starfish’s published private-market totals and next to any prime with a servicing line. GEO in 2027 now has to carry both the commercial thesis and the reputation repair.

Attitude control and propellant margin

LINK lost two of three reaction wheels and burned extra fuel holding attitude. That is now the company’s public reliability mark. Any NEXUS design review that does not show a different control and propulsion stack will be read against this flight.

Mission-objective abort

NASA Release 26-067 is unambiguous: no capture, no boost. Residual RPO can still produce data. It cannot be briefed as a rescue.

Customer concentration

The largest disclosed award is one NASA Phase III. SpaceWERX and DIU/SDA work is real and smaller or unpriced at the company level. A second civil or national-security flight award is the concentration test.

Capital and valuation opacity

Post-money on the $12 million raise is unpublished. Form D sold $880,635 against a $1.75 million 2024 offering. Aggregator totals disagree with each other. Treat cap-table claims outside those two primary/company figures as unverified.

Acquisition integration

Atomos brought a plant, a team and a 2024 demonstration with problems. Quark heritage is useful only if Broomfield can turn NEXUS hardware without repeating that commissioning record.

Schedule risk on GEO

Ariane 6 in the second half of 2027 is a company date. DaaS Phase II is a government option. Either can slip independently of the other.

Arizona LLC articles name Ghonhee Lee, Kaleb Forrest Beebout and Nicholas Peter Liapis as members from late 2019. Beebout and Liapis are consistently described as co-founders in secondary directories. Current operating titles for Beebout and Liapis are not confirmed on the company newsroom as of this refresh and are not printed as active executive cards.

No current statutory board roster is published on the company site. The 2024 Form D for Katalyst Space Holdings, Inc. lists Ghonhee Lee as executive officer. Treat governance as founder-led unless a later filing says otherwise. Product-team cards are not officers.

Katalyst Space Technologies, LLC is a private operating company. It files no 10-K, 10-Q, 8-K or DEF 14A. The public record that does exist:

Katalyst Space Technologies, LLC was organized in Arizona on 13 November 2019. Company channels often say 2020. The founding group came out of guidance, navigation and control work rather than a prime-contractor bus line. Early revenue was Air Force SBIR money on object characterization, a cislunar observation network concept and a retrofittable awareness package later marketed as SIGHT.

In March 2025 the company closed the acquisition of Atomos Space and announced it on 24 April. That deal bought a Broomfield integration plant, an orbital-transfer-vehicle design and people who had already flown a difficult 2024 demonstration. In September 2025 NASA awarded Katalyst about $30 million to turn a planned servicing demonstrator into a rescue ship for Swift. LINK launched from Kwajalein on 3 July 2026. On 19 August NASA and the company ended the capture-and-boost attempt.

The same summer the company raised $12 million for a GEO vehicle called NEXUS, booked an Ariane 6 path for 2027 and took a seat on the DIU/SDA deorbit study next to Firefly and D-Orbit USA. That is the entire arc: a Flagstaff LLC that used SBIR work and an acquired Colorado factory to get a robot on orbit inside a year, then learned in public that reaching the target and grabbing it are different jobs.