I fear the day that a UAV has the payload capacity available to lift a "minicomputer." Also not to mention what is a pretty impressive feat; being able to have code that can perform all the tasks (quickly), mentioned in the summary, running on a minicomputer. Never mind the ability to lift one.
In all seriousness, google is charging their "customer" $300 for this service for 10Yrs, and they are then charging their true customer for the ability to know all the online activity of the "customer." It is like Google is reading right from the cable companies play book.
I am having a hard time with how we get things that were not meant to be interchangeable to link up in orbit to take on a new mission. Now if we could build a WALL-E (SPACE-E) that is capable of catching, manipulating, disassembling and construction of a new satellite that would be pretty impressive. However by the time we develop that technology we will have spent a considerable amount of money.
Perhaps the path forward is to develop satellites that can either connect (dock if you will) to the ones currently in orbit and use some of the preexisting equipment on board to perform a new mission, or build new satellites with a "SPACE-Port" that allows future satellites to dock with it and use the host's equipment. Right now I see the problem on the order of trying to connect a CDC 6060 with a K Computer.