Edge Computing in Space: Necessity or Luxury?

The increasing reliance on data for decision-making and the rising number of exploration missions, and EO constellations with high resolution sensors with increased revisit frequencies is growing data generated onboard satellite exponentially. However, the bottleneck on the ground to handle data poses a challenge to obtain low latency actionable insights for critical decision making. Thus, processing data near to the sensor is seen as a suitable solution for faster delivery and reducing the data volumes to be downlinked, and as a result, reducing the cost of data transport. But the associated costs for space-based edge computing solutions come at premium costs. The question that arises then is if these services are necessary or luxury items. 

Existing demand for in-orbit cloud services is focused on in-orbit storage solutions. NSR’s Cloud Computing via Satellite, 3rd Edition report forecasts over $100 Million in cumulative revenues to be generated in the next decade for this very nascent market, mainly for storage via orbital infrastructure.  

Space players targeting the storage segment are also offering computing solutions and are currently in development phases to provide services from 2023-2024. The difficulty to access space-based assets physically coupled with additional security layers, could make it an attractive solution for Gov/Mil and Financial institutions. These customers are ready to pay a premium for highly secured solutions as they are very aware of their exposure to cyber-attacks.  



As the market is currently in its embryonic stages, the ability to provide the services first will be an advantage for these service providers. Despite lower total opportunity size for storage solutions, the corresponding edge computing service revenues are expected to be many folds larger. This demand is expected to be driven by the need for higher operational efficiency, low latency and hence processing part of data closer to the source. However, the demand for edge computing will materialize only during the second half of the decade once the reliability of these services is tested and proven with better customer awareness. EO, lunar/planetary and science missions will drive the early adoption of edge computing solutions in space.

Early Adoption by EO

Edge computing and on-board processing enables the following functions for EO applications:

• ‘Smart’ discarding of images (pre-filtering and discarding of corrupted and cloud-covered satellite images)


• Rapid extraction of information using AI for use-cases such as disaster monitoring, environmental monitoring, illegal fishing


• Decentralising the data processing and increasing data reliability

Players such as Unibap have successfully tested on-board processing on satellite to provide low latency information products. And more recently, operators like Spire have launched satellites with supercomputers onboard that offer computing power of up to 1 Teraflop in-orbit.

However, on-board processing may not be a suitable solution for all operators as it needs additional hardware capabilities on the satellite that increases mass and power budgets, in turn increasing cost of deployment to orbit. Also, satellites in-orbit that do not support on-board processing hardware cannot do any local processing. Thus, there is a need for satellites in orbit to handle compute operations, and even intensive AI/ML models. Companies such as SpaceBelt and LeoCloud are trying to bridge this gap.

Edge computing enables a reduction of data volume to be downlinked, leading to lower cost of data downlink and low latency. However, these services come at premium costs compared to data downlink to ground station. Furthermore, with improving ground capabilities in hardware and software for data processing and analysis, the question ‘is this trade-off worth it?’ arises.

With multiple operators still focusing on deploying their constellations and acquiring imagery, these services will be deployed in subsequent phases. In the short- and medium-term, edge computing services in space would seem more of luxury. But these will also provide competitive advantage to the operators, especially those that cater to Gov/Mil customers.

Edge Computing Beyond Earth’s Orbit

Future missions to the Moon, Mars, and beyond require faster decision-making, such as suitability of the environment for conducting experiments on the surface, advisable medical treatments, equipment status and suitability for EVAs, etc. Existing infrastructure poses a challenge in such scenarios due to significant delays in data transmission to Earth and back. For example, it takes about 5-20 minutes to receive a radio signal from Mars to Earth, leveraging the Mars orbiter as a relay at UHF frequency, and much longer to acquire high-resolution images and spectral data. Hence, processing data closer to the source leveraging edge node will reduce the time to make decisions and reduce the volume of data to be transmitted back to Earth.

Edge devices will utilize data from images, gasses, systems health data, etc., and run AI/ML algorithms to produce ‘Yes’ or ‘No’ decisions in a matter of minutes or seconds. The biggest challenge for such computations to date has been the scarcity of processing power, physical space, and temperature regulation of the spacecraft. The availability of radiation-hardened high-power processors and changing approach to designing ML models with available compute resources reduces the barrier to edge computing.

It will further enable use-cases such as crew health monitoring and facilities monitoring onboard ISS and commercial space stations at a much faster pace. Exploration missions will be able to reduce data transmitted to Earth and increase data quality. Software companies like IBM have collaborated with HPE and the ISS National Lab to demonstrate edge computing utilized for DNA sequencing project onboard ISS.

The surge in the number of exploratory missions, planned crewed lunar missions and habitats in the long term will lead ever greater volumes of scientific data. Thus, the need to quickly process this data will make a compelling case for setting up infrastructure on the surface to process data locally. As an example, Lunar data centers with storage and edge capabilities are in the planning stages for implementations towards the end of the decade to help the development of critical moon missions (i.e., Rovers, orbital platforms, surface habitats).

Bottom Line

The edge computing in space market is at very early stages and will slowly ramp up to an accelerated growth phase towards the second half of the decade. The needs are there for low latency processed data, faster data delivery and lower data transport cost, which will all add up to drive the adoption of edge computing solutions.

The proliferation of data generated from constellations, their increasing number of satellites, the growth in sensor capabilities and the need for faster data delivery will all combine to spur satellite on-board processing for EO applications. The foreseen growth in data volumes will necessitate the deployment and use of edge computing solutions to address these customer needs.

And if we add the coming surge in space exploration and crewed missions to orbit, Moon, Mars and beyond, the corresponding data volumes will make edge computing a necessity longer term to meet mission objectives.