Processing satellite data before transmission gives space computing a practical commercial role. Operators can select useful imagery, identify events and send compact results to Earth. Recent hardware deployments and a new supply contract show several routes to participate, from onboard computers to application software. SignalSpore sees the clearest commercial case where customers need a small, timely result from a much larger stream of observations.
More useful information from each contact window
A satellite’s communications capacity has to serve everything it needs to send home. If onboard software can reduce the volume required for a customer’s task, the same link can carry more useful results. An operator could prioritize an event alert, transmit selected images or use the freed capacity for another customer’s observations.
There is flight evidence behind the approach. ESA reported successful cloud screening aboard PhiSat-1 in 2020. More recently, the NAVI-Orbital team reported running a vision-language model on Loft Orbital’s YAM-9 satellite, including two newly acquired images in April 2026. Its workflow uses text descriptions to help operators select imagery for retrieval. Broader validation remains a step toward routine service.[1][2]
Smaller payloads free transmission capacity
Illustrative model • 1 TB of initial data • 100 Mbps effective payload throughput
Source: SignalSpore calculations. Decimal units. Transmission time = transmitted bytes × 8 ÷ payload bits per second. Percentages are assumptions, not measured mission results. Time is accumulated link use, excluding processing and waits for connectivity; it is not elapsed delivery time.
In the 10% case, transmitting 100 GB takes 2.2 hours, freeing 20 hours of link use against the full 1 TB. The opportunity is valuable when the selected output preserves what the customer needs. Savings depend on processing cost, quality and communications pricing; a 90% reduction in bytes does not imply a 90% reduction in mission cost.
Commercial evidence is broadening
October announcements show progress in both deployment and procurement. Loft Orbital and Orbitworks report that the first spacecraft in their planned 50-satellite constellation carries two NVIDIA Jetson AGX Orin modules. Their stated application areas include maritime monitoring and disaster response, with third-party software intended to run on shared spacecraft. The first satellite is in orbit; the broader constellation remains a rollout plan.[3]
Unibap separately announced its first contract to supply satellite edge-computing platforms for a signal-intelligence mission. The initial order covers a handful of computers for a confidential European space-defence customer. Its value and the chosen platform were not disclosed. The order provides a concrete example of mission-specific demand for onboard processing.[4]
Several parts of the supply chain can participate
Selected examples • Commercial implications are SignalSpore analysis
| Market layer | Evidence | Potential commercial opportunity |
|---|---|---|
| Onboard computers | Unibap reports an initial satellite-computer supply contract.[4] | Equipment supply, integration and follow-on orders as customers add missions. |
| Shared spacecraft | Orbitworks and Loft report two Orin modules on the first constellation satellite.[3] | Hosting application workloads on common infrastructure. |
| Applications and validation | NAVI-Orbital reports image interpretation aboard YAM-9.[2] | Detection software, mission-specific validation and model maintenance. |
A clearer route to repeat business
The opportunity extends beyond the processor. An operator needs software that recognizes the right event, works within the spacecraft’s power budget and delivers output the customer can trust. Suppliers that combine these capabilities can address a defined operational requirement and build experience that carries into subsequent missions.
For application developers, shared spacecraft could lower the infrastructure commitment needed to enter the market. Maritime alerts, fire detection and selected-image delivery each provide a specific service to sell. Repeat revenue would depend on customer adoption, reliable delivery and commercial terms. Disclosures of pricing and repeat orders would make it possible to estimate revenue per spacecraft and build better grounded forecasts.
The next useful commercial signals are follow-on equipment orders, paid application deployments and customer renewals. Measured delivery times and detection quality would help buyers compare services. Each successful customer deployment can also produce a reference customer and tested integration work, giving suppliers a firmer starting point for the next contract.
Sources
- ESA, PhiSat-1 onboard image processing, 22 October 2020.
- Delfa Victoria et al., NAVI-Orbital, arXiv:2606.18271v1, 2026. Research preprint.
- Loft Orbital, Orbitworks constellation announcement, 12 October 2026.
- Unibap, first space-based signal-intelligence order, 13 October 2026. Company release.