Google’s “Project Suncatcher” is built on two “free” concepts: “free” (as in, unlimited) solar energy, and “free-space” (as in, wireless) optical data. This combination of “unlimited” power and “high-bandwidth” data is what makes the orbital datacentre possible.
The “free” energy comes from solar panels that are 8-times more productive in orbit, providing an “unlimited, low-cost” power source for the “Google TPUs.” This solves the energy-and-carbon problem of the $3 trillion terrestrial industry.
The “free-space” data comes from “optical links.” These lasers will “connect” the 80 satellites in a constellation and “beam their results back” to Earth. This is the “high-bandwidth” nervous system that allows the distributed AI “brain” to function.
This is the core “tech stack” of the orbital AI. However, Google’s “cautionary note” admits that both parts are “significant engineering challenges.” The solar panels must power a “thermal management” system to cool the TPUs, and the “free-space” optical links must be perfectly reliable.
The 2027 prototypes will be the first “milestone” in testing this tech. Google must prove it can master both “free” energy and “free-space” data before this “moonshot” can become a reality.
