At AWS re:Invent 2022, AWS announced the results of a successful experimental spacecraft running AWS compute and ML software on an orbiting satellite. The stated purpose of the experiment was to enable orbiting satellites to process data directly onboard satellites using the cloud. This would enable the satellite to collect massive amounts of data, analyze it in real time, and downlink only a subset of the most useful images for storage and further analysis.

This experiment also has massive implications outside of the scope of what was shown. Because the satellites are running AWS compute, a fleet of them could act as a network of interconnected machines, similar to terrestrial AWS data centers. Satellite operators would then benefit by ride sharing between machines, enabling more powerful computing than has been seen in space. The satellites could also act as proxies for ground stations, increasing the downlink capability of all satellites in the fleet.

From a security perspective, there is some concern about running distribute computing software in a new environment. Errors in the data on one satellite could poison the data replicated to other satellites. In addition, and satellite interfacing with this fleet could potentially run malicious software if attackers found a way to escape the AWS compute partition and access the underlying hardware.