Welding in Orbit: Momentus, ARCSPACE and the Business of Building in Space

$2.6 billion is where the in-space servicing, assembly and manufacturing market is expected to land in 2026, up from $2.18 billion a year earlier. Nearly every dollar of it will eventually depend on an unglamorous mechanical act: joining two pieces of metal so the joint survives vacuum, thermal swings of hundreds of degrees, and a decade of vibration.

On October 15, Momentus signed a contract to test exactly that. ARCSPACE, a French aerospace startup, will fly an autonomous welding payload on Momentus's Vigoride-9 orbital service vehicle. Launch is projected for 2027.

A small market carrying a very large load

ISAM is three businesses wearing one acronym. Servicing means repairing, refueling or relocating satellites already in orbit. Assembly means building large structures once they are up there. Manufacturing means making things in microgravity and, often, bringing them back. All three sit at very different stages of maturity, and lumping them together hides how uneven that progress is.

Together they generated $2.18 billion in 2025 and are tracking toward $2.6 billion in 2026, roughly 19% growth. That is a rounding error next to launch and satellites. It is also the layer that decides whether anything large ever gets built in orbit. The wider space economy runs from $630 billion in 2023 toward a projected $1.8 trillion by 2035, according to World Economic Forum figures cited in the same market research.

The distance between those figures and the sector's actual output is the story. Space spent a decade getting good at putting things up. It has spent almost no time getting good at putting things together once they are there.

Welding is the hard part nobody films

On Earth, gravity does half the work of a weld. It shapes the molten pool, helps trapped gas rise out of the metal, and gives the operator a stable reference. In low Earth orbit, none of that holds. Surface tension dominates, bubbles stay where they form, and cooling behaves in ways the textbooks handle as edge cases.

Autonomy is the other half of the problem. Round-trip latency to a ground controller rules out joystick welding. The payload has to decide, in real time, whether the seam it just laid is sound. ARCSPACE's payload will run a series of autonomous welding operations on Vigoride-9 and transmit operational data — weld quality, process stability, operational performance — back to engineers on the ground for analysis.

Demonstrating autonomous welding in orbit is a critical step toward building larger and more capable structures in space.— Guillaume Mohara, chief executive, ARCSPACE

That is the real deliverable: not a welded bracket, but a dataset about how metal behaves when gravity is switched off. Whoever owns that dataset owns the design rules for the next generation of orbital structures.

Servicing got there first, and mostly without welding

The servicing half of ISAM is a year or two ahead, and it is the part with government money behind it. On July 21, DARPA and SpaceLogistics launched the Mission Robotic Vehicle to geostationary orbit — the first privately owned, operational in-space servicing mission, built around a robotic payload developed with the U.S. Naval Research Laboratory. It carries a 10-year-plus operational life and an obligation to share servicing data with the U.S. Space Force.

Starfish Space has pushed the same idea into procurement. In February it won a $54.5 million Space Force contract for a dedicated Otter servicing vehicle, its second Otter purchased by the service under a program meant to buy capability before it is fully proven. That is a meaningful shift: the customer is buying the mission, not the research.

The cautionary case sits right next to the successes. Katalyst Space's link-up with NASA's Swift telescope — the first attempt to have a commercial vehicle physically capture and push an unprepared government science satellite — ended with NASA abandoning the reboost and the Link vehicle reentering the atmosphere. Proving a servicing concept on orbit and turning it into a repeatable service are two different problems.

Maturity assessment based on launched missions and announced 2026 contracts

Welding sits at the seam between assembly and manufacturing. It is the first tool you need to build a structure bigger than a single launch fairing, which is why a welding demo carries more weight than its small contract value suggests.

Momentus bought its way back into the conversation

Momentus is the more interesting party to this deal, because it is the one that already went through the fire. Founded in 2017, it raised more than $153.8 million and carried a valuation north of $1 billion after going public. Then the arithmetic caught up: a cancelled mission, a cash crunch, and a Nasdaq delisting notice. By late 2024 it was surviving on a $2.75 million private placement.

The recovery came from narrowing the ambition. Instead of selling transport to customers who might not have payloads, Momentus resold its Vigoride vehicle as a hosted-payload platform — power, communications, command and control, and operational staff, without the customer having to build or fly a spacecraft. ARCSPACE is the latest company to buy that shortcut.

As we wrote in October, the on-orbit servicing arms market is still fighting a brutal mismatch between a $310,000 seed round and the $30 million a life-extension mission can cost. Hosting someone else's experiment is a cheaper way to stay in the game while the real market arrives.

The economics still do not close

The cost of getting mass to orbit is the number that shapes every one of these businesses. On a SpaceX Falcon 9 rideshare, it runs around $7,000 per kilogram. A startup planning orbital AI data centers put the problem bluntly earlier this year: at that price, launch economics do not yet work for compute.

Manufacturing in orbit has an escape hatch that compute does not. It only has to work for products that are worth far more than their launch cost. Varda Space Industries found that market in pharmaceuticals: it grows drug crystals in microgravity, returns them through the atmosphere, and has now signed an agreement covering twenty further capsule returns after flying its sixth mission.

Welding inverts the calculation entirely. If a structure is assembled on orbit from material already up there, you never pay the launch cost for the finished object. The catch is that today there is almost nothing up there to build with, which is why every serious in-space assembly plan starts with a supply chain that does not exist yet.

The first customers are not commercial

Almost every ISAM business plan points at a commercial market that has not arrived. In the meantime, the customers signing checks are governments, and that reality shapes what gets built and when.

DARPA's servicing program exists for a specific reason: repositioning and repairing satellites extends the life of assets the U.S. government already paid for. The Space Force's Servicing, Mobility and Logistics portfolio is the intended long-term home for that capability, and the RSGS robotic payload is designed to be handed to it after the initial demonstrations. The spending is justified as resilience, then reused for commercial work once the technology matures.

NASA is the other anchor customer, and its position has been unstable. The agency spent years encouraging commercial low Earth orbit destinations, then moved to step back from backing standalone stations, then reversed course in June 2026 after sustained industry pushback. A Government Accountability Office review running alongside that reversal flagged the underlying risk plainly: commercial stations may arrive later than planned, leaving a gap in U.S. presence in low Earth orbit.

That gap is where demand for assembly and manufacturing eventually lives. A commercial station needs structures it cannot fit inside a single launch — trusses, radiator arrays, modules joined end to end. Welding and robotic assembly are the tools that turn a sequence of launches into one building.

The practical effect is that orbital construction is being funded today as national-security and civil infrastructure rather than as a standalone business. That is how launch, GPS and weather satellites all started. It also means the commercial tipping point depends on government demand holding long enough for unit prices to fall — a bet that has paid off before and is not guaranteed to pay off again.

What to watch next

The welding contract is small. The direction it points in is not. Every serious plan for orbital platforms, fuel depots and large solar arrays assumes that metal can be joined reliably in a vacuum by a machine nobody is steering. Momentus and ARCSPACE are trying to prove the smallest version of that claim before anyone has to bet a station on it.