More than $300 billion worth of satellites are estimated to be in the geosynchronous orbit. Many of these satellites have been retired due to normal end of useful life, obsolescence or failure; yet many still have valuable components, such as antennas, that could last much longer than the life of the satellite. [...] One of the primary drivers of the high launch costs is the weight and volume of antennas. The repurposing of existing, retired antennas from the graveyard represents a potential for significant cost savings.
"Satellites in GEO are not designed to be disassembled or repaired, so it's not a matter of simply removing some nuts and bolts," said David Barnhart, DARPA program manager. "This requires new remote imaging and robotics technology and special tools to grip, cut, and modify complex systems, since existing joints are usually molded or welded. Another challenge is developing new remote operating procedures to hold two parts together so a third robotic `hand' can join them with a third part, such as a fastener, all in zero gravity. [...]
Phoenix specifically seeks technologies for developing a new class of small "satlets," or nanosatellites, which can be sent more economically to the GEO region through existing ride-along services with commercial satellite launches and then robotically attached to the antenna of a nonfunctional cooperating satellite to essentially create a new space system.
This has been making the rounds lately with a lot of people reposting it as if this is a thing that is going to happen, but DARPA puts out these kinds of proposals all the time, sometimes for seriously wacky, probably-impossible things. What they're saying is, "Here's something we'd like, and if you figure out how to build it, we'll probably buy it."
Neat idea, though.