Six weeks before a satellite ships, the integration lab is often still waiting for hardware. A reaction wheel sits in customs. The power-control unit is on its third revision. The flight-software team, hired eighteen months earlier, has a test bench wired to almost nothing.
That gap used to be absorbed by schedule. A launch date slipped, a customer grumbled, a new window opened. Rideshare slots and constellation build plans have tightened the math. Hardware that arrives late does not move the launch; it compresses the months of integration and testing into whatever is left.
Adam Chikha and Serge Barral spent years on the other side of that problem. At Airbus they built and maintained the system-level simulators that let engineers test how a spacecraft's software, components and failure modes interact before the real parts were bolted together. Those environments took decades and large teams to mature. A New Space startup with forty engineers and a Series A cannot reproduce them.
The test bench that arrives before the hardware
Asynchronics, the Warsaw company the two founded in 2023, is built on a narrow claim: the expensive part of spacecraft verification is not the physics, it is the plumbing. Every mission rebuilds models of the same reaction wheels, sun sensors and power units, translating supplier documentation into yet another bespoke simulator. The work is duplicated across the industry and rarely reused.
On November 17 the company announced a $3.1 million seed round led by b2venture, with Movens Capital, Heartfelt and Hard2beat participating. The money is earmarked for a catalogue of reusable component models, a set of tools for systems engineers, and a team that is supposed to grow from about eleven people to twenty-four, with an office in Toulouse.
What the primes built, and what New Space could not
Large integrators have run high-fidelity simulators since before private spaceflight existed. Airbus, Thales and their suppliers wrote them in C++, staffed them with specialists, and treated them as capital equipment. The model is expensive but it works: by the time hardware shows up, the software has already been tested against something that behaves like it.
New Space broke the economics of that model without replacing the need. “Smaller companies or mid-sized companies do not have access to this kind of simulation tooling,” Chikha told Payload. “What we did was we took our vast experience, and created something better.”
The failure mode is not that small teams skip testing. They test constantly, just later and more manually than they would like. The bill arrives as rework: a component interface that behaves differently from its datasheet, discovered during thermal-vacuum, with the launch slot already paid for. Asynchronics cites a familiar industry figure, that 43% of small satellites fail. The company uses it as marketing, and the exact denominator is debatable, but the pattern it points at is real.
An engine given away, a catalogue sold
The engine underneath the company is NeXosim, an open-source discrete-event simulation framework written in Rust. It reached version 1.0 in February 2026 and is distributed under the MIT and Apache licences. Models are actors with typed inputs and outputs; the runtime schedules them across CPU cores and handles synchronization, which is why the company can claim three to five times the speed of legacy tools. That number is a vendor benchmark, not an independent one.
Open source is doing commercial work here. A verification tool is only useful if engineers can inspect its assumptions, and b2venture's investment note makes the same argument: teams need to see where a simulation ends and a guess begins. Giving the engine away also makes the models portable, which matters for the second half of the business.
That second half is NeXosim.io, a marketplace of manufacturer-validated digital twins. Each model is built or approved by the company that makes the hardware and licensed per user per year. Revolv Space lists its SARA solar-array rotary actuator at €24,000 per user per year; NewSpace Systems lists the Aquila NFSS-411 digital sun sensor at €6,000. Buyers add a model to a Rust project the way they would add any dependency, then run a full mission bench against it.
The pitch to hardware makers is recurring revenue from software they already have to document. The pitch to integrators is less non-recurring engineering: stop reverse-engineering every subsystem, and find integration problems before the units arrive. Asynchronics says a NewSpace Systems reaction wheel twin, tested with the same procedures used on the real unit, produced identical readings. That is one component on one bench, and the company's own account of it, but it is the kind of evidence the market will demand.
Fifty twins in twenty-four months
The plan is to reach at least fifty digital twins within two years, concentrated on the most common parts: reaction wheels, solar-array drive assemblies, power-control-and-distribution units. The company is also building an AI-assisted model-generation tool that it says can cut twin creation from months to days, plus visual tooling so engineers can see the system they are assembling.
Set against the rest of November, $3.1 million is a rounding error. Ramon.Space raised $124 million for orbital computing, Dcubed raised €27.2 million for actuators and solar arrays, and CesiumAstro closed a $730 million Series D. Most of that capital is aimed at building more hardware, faster. Asynchronics is betting that the constraint moves to proving the hardware works, and that verification capacity has to scale with production capacity.
What has to be true
Three things, none of them settled by a seed round.
The first is trust. A digital twin that says the system is nominal, followed by a failure on orbit, turns a simulation vendor into a liability question. Manufacturer validation helps, but validation has a scope, and the scope will be argued about after the first bad outcome.
The second is catalogue breadth. Fifty twins covers the popular parts and little else. Missions are built from long tails of custom and legacy hardware, and a marketplace that cannot model the odd component still leaves the integrator writing code.
The third is competition. Siemens sells simulation through Xcelerator, and ESA maintains SIMULUS for its own ecosystem. The Asynchronics founders argue SIMULUS is not truly open source and that companies with subsidiaries in Japan or the United States cannot always access it. If that access gap closes, the open-source advantage narrows.
What the round does buy is time to test the central assumption: that a validated model can be sold like a component, by the company that built the component, to every team that uses it.
If that holds, the digital twin stops being a project deliverable and becomes a line item in the bill of materials. If it does not, Asynchronics is a useful open-source engine with a small paid catalogue, which is a respectable software company and a much smaller business.