FRONTIER SIGNAL | DEEP DIVE | SPACE INFRASTRUCTURE & INVESTING | 9 MIN READ | SATURDAY, 12 SEPTEMBER 2026
KEY TAKEAWAYS
Nine minutes. The product is not the weld.
ArcSpace plans a 12 kilogram orbital demo in 2027 and targets commercial availability from 2028.
Vacuum suits electron beams. Microgravity complicates positioning, heat rejection and contamination.
The first autonomous space weld happened in 2024. ArcSpace must prove compact, repeatable operation.
The investable layer is robotics, inspection, power, host vehicles and operations.
Decisive evidence means metallurgy, defect data, process control and a paying customer.
A servicing spacecraft reaches a dead satellite with valuable hardware and no grapple fixture. It can photograph the target, perhaps push it, but it cannot safely attach. The satellite was built for launch, not repair. ArcSpace wants to fix that with a move familiar on any factory floor: weld a connection straight onto the structure.
The French startup says a 12 kilogram autonomous electron-beam payload will fly on a host spacecraft in 2027, each test run including more than 100 seconds of short welds. It has raised more than EUR 2 million and targets commercial availability from 2028.
This is not a first-weld-in-space story. ThinkOrbital and TWI produced the first autonomous electron-beam weld in space in May 2024, about one centimetre across, and returned the samples. ArcSpace's proposition is narrower and possibly more useful: can welding become compact and repeatable enough to ride as a hosted payload rather than a bespoke mission?
THE NUMBERS
12 kg planned payload · 2027 orbital demo · over EUR 2m raised · 100+ sec welding per test · 1 cm prior weld, 2024
ArcSpace figures are company-reported plans, not flight results. The one centimetre weld was ThinkOrbital and TWI's separate 2024 test, on a Falcon 9 first stage.
What happened
Founded in 2022 by Guillaume Mohara and Adam Abdin, ArcSpace has spent roughly 18 months on use cases with about 30 space companies. The same tool welds or cuts by changing parameters.
The first commercial idea is pragmatic. A servicer could weld a capture plate onto a satellite that was never designed to be grabbed, which is exactly the problem the standard docking port was invented to solve, approached from the opposite direction.
Named interest is demand signal, not contract. ArcSpace reports talks with Airbus, Thales and Mitsubishi, plus work with Exotrail and Infinite Orbits.
2027 is a qualification step. A host lowers launch cost and allows repeated runs across materials. The host, the launch slot and the qualification plan are all still missing.
How electron-beam welding works in space
An electron gun fires a focused beam at metal, and that energy becomes heat which melts the edges of a joint. Air scatters electrons, so on Earth this happens inside a vacuum chamber. Space supplies the vacuum free, and allows narrow, deep welds with no shielding-gas bottle.
The convenience ends there. In microgravity molten metal does not settle as on Earth, and surface tension dominates the weld pool.
Vacuum removes convective cooling, so heat must radiate away through the structure, and metal vapour can contaminate optics or arrays.
Two pieces still need aligning and holding while the spacecraft moves. The weld head is one part. Arms, clamps, cameras and metrology make the joint.
The real product is a trusted joint
A bead proves the tool can deliver energy. A bankable joint needs controlled geometry, repeatable penetration, defect detection, traceable process data and an acceptance standard.
On Earth an inspector has radiography and ultrasonics. In orbit you need machine vision during the weld, then remote evaluation.
Risk tolerance is not uniform. High for a capture plate on a dead satellite, far lower for a pressurized habitat or a load-bearing truss.
So the market arrives in stages, and qualification becomes a product in itself. Whoever owns the process data may hold more value than whoever supplies the beam.
Why it matters commercially
Most spacecraft are still launched as finished objects.
The rocket decides the shape. Antennas, arrays and structures have to fit a fairing and survive launch loads, a constraint we followed all the way to a satellite flattened into the shape of a dinner plate.
Reliable joining in orbit would loosen that: launch smaller modules, assemble larger systems, repair assets rather than replace them, and reach satellites with no grappling interface.
The cautionary precedent is OSAM-1. NASA killed the servicing project in 2024 after more than $2 billion, citing technical, cost and schedule problems and a market moving away from refueling unprepared spacecraft. The need can be real while the economics are not.
THE OTHER WAY TO SOLVE THIS
If the hard part is joining metal in orbit, the alternative is to never join it. We wrote about NASA building a telescope the size of a city out of six shoeboxes, holding formation instead of structure. Both accept the fairing as the enemy. They disagree on whether the answer is a weld or a flight computer. Formation cannot carry a load.
Where the money is moving
The seed round is over EUR 2 million, led by DEPO Ventures, with SCE Freiraum Ventures, IRDI Capital and angels.
The EU-backed TAC project is a 30-month, EUR 4.0 million program for welding-based satellite capture. ArcSpace is one consortium member, so that is not funding to the startup.
The pattern is early-stage and mission-led: small rounds, agency program, partnerships with servicers. The same sequence we traced when pharmaceutical mini-factories started flying to orbit. Capital funds flight evidence long before it funds factories.
Listed company read throughs
No listed company's earnings depend on orbital welding. These are exposure to the layer it needs, and the demand side is the part we sized when the race to replace the space station went private.
Redwire, NYSE RDW. The clearest public read-through to in-space infrastructure. Q2 2026 revenue $117.1 million, up 89.6 percent, backlog $542.1 million, net loss $41.0 million. No disclosed ArcSpace link, and welding is not an earnings driver.
MDA Space, TSX and NYSE MDA. Relevant for manipulation, not welding. Robotics and space operations made C$99.5 million in Q2 against C$4.0 billion backlog. A weld head needs an arm to steady the joint. Canadarm3 drives the valuation, not this.
Voyager Technologies, NYSE VOYG. Demand side, via Starlab. Q2 2026 revenue $52.7 million, backlog $335.5 million, adjusted EBITDA negative $37.5 million. The nearer question is whether station schedules and funding hold.
Private company watchlist
ArcSpace must turn customer discovery into flight data, then a paid service. Watch for a named host, a launch contract, and a funded 2028 target.
ThinkOrbital and CisLunar Industries are the comparator. ThinkOrbital led the 2024 weld, CisLunar supplied the 20 kilovolt power system. That raises the bar for every later claim.
Infinite Orbits raised EUR 40 million in 2025 for inspection and life-extension work, and ArcSpace reports commercial work with it. A servicer putting a weld head on a revenue mission would beat another lab milestone, the way a robot launched to catch a falling NASA telescope beat a decade of servicing studies.
The honest catch
ArcSpace has not welded in orbit. The payload must survive launch, avoid contaminating its host and hold quality across thermal conditions.
A flat coupon is not a joint. It still has to show how a robot prepares, aligns and inspects real hardware.
Adoption will be slow. Owners are conservative, insurers want evidence, standards are young.
Welding creates permanence. A bad bolt can be swapped. A bad weld may demand cutting and rework. Removing fasteners does not remove fatigue, defects or thermal-cycle risk.
OUR VIEW
ArcSpace is an important technology signal, not yet an investment catalyst. The 2027 mission tests whether an industrial process can shrink into a compact orbital tool. It does not settle whether customers will trust the result.
The near-term opportunity is the whole workflow rather than the beam: manoeuvre, grasp, align, weld, inspect, certify. Robotics, station infrastructure and mission integration may capture value long before welding revenue is material, and ArcSpace only matters commercially when a customer pays for an outcome rather than a demonstration.
THE ONE QUESTION EVERY INVESTOR SHOULD ASK
Who inspects the joint, against which standard, and who carries the liability if it fails? Until all three have answers, this is a tool, not a service.
Five signals to watch next
A named flight and host. A contracted launch and integration partner cuts schedule risk.
Post-flight metallurgy. Penetration, porosity, cracking and strength across samples.
Closed-loop control. Sensing that adjusts beam power, focus or travel speed in real time.
Inspection in orbit. Remote non-destructive evaluation gates the higher-value joints.
A paid mission, plus standards and insurance. A customer buying capture or repair rather than test time, and an insurer accepting a qualification rule.
Quick questions
What is orbital welding?
Joining metal in space. ArcSpace plans a focused electron beam, suited to vacuum.
Has welding already been done in space?
Yes. Crewed experiments date to 1969, and ThinkOrbital with TWI made the first autonomous in-space electron-beam weld in 2024.
What is new about ArcSpace's approach?
A 12 kilogram hosted payload running repeated autonomous tests, aimed at a commercial tool rather than one experiment.
Why would satellites need welding?
To create attachment points on unprepared satellites, support repairs, cut damaged hardware and join structures too big for one fairing.
No, ArcSpace is private. Listed exposure is indirect, through robotics, infrastructure and mission platforms. None is a pure orbital-welding play.
READ MORE: SPACE INFRASTRUCTURE SIGNALS
Four pieces on building and repairing in orbit.
The Standard Port That Could End the Era of Disposable Satellites 21 Aug. The interface problem, by design.
NASA's Telescope Is Falling. A Robot Just Launched to Catch It. 30 Jun. Catching what was not built to be caught.
Who Owns Orbit? The ISS Is Retiring, and the Race to Replace It Just Went Private 10 Jul. Who would buy orbital assembly.
The Satellite Shaped Like a Dinner Plate 31 Aug. What the fairing does to design.
Sources
European Spaceflight on the ArcSpace demo: payload, flight, test runs
The French Tech Journal on the round: funding and use case
EU-Startups on the raise: investor names
ESA, space-made weld scrutinised: the sample, analysed
NASA, OSAM-1: mission record
SpaceNews on the OSAM-1 cancellation: cost and reasons
Planned demonstrations, company statements and funding announcements do not establish technical or commercial success. This is general information, not investment advice.