Google Unveils Project Suncatcher: Space Solar-Powered AI Data Centers

Google has unveiled plans to deploy AI data centers in space powered by solar energy to secure massive amounts of power.

Google announced on the 4th, local time, its “Project Suncatcher,” which aims to build data centers in space that directly utilize solar energy. As AI demand surges, the power consumption of data centers performing computations is growing exponentially. Google stated, “The sun emits energy exceeding 100 trillion times the total global power production,” adding, “In the future, space could become an ideal location for expanding AI computing.”

The basic concept involves launching multiple small satellites equipped with large solar panels and Google’s proprietary AI chips, the “Tensor Processing Unit (TPU),” into low Earth orbit. Google explained, “Solar panels in the right orbit can generate energy up to 8 times more efficiently than on Earth.” The satellites will form an AI computing cluster through optical communication between them.

A key challenge is accelerating communication speeds between satellites, as hundreds to thousands of TPUs must exchange computational results in real time. Google adopted a method of flying satellites in extremely close formations, spaced hundreds to thousands of meters apart, to enhance communication speeds. The company’s research team achieved bidirectional speeds of 1.6 Tbps (terabits per second) in ground tests. For the space cluster, a minimum of 10 Tbps is required. Additionally, Google’s Trillium TPU maintained operation without critical damage in a radiation environment equivalent to a 5-year mission in space.

Google revealed plans to collaborate with the private space company Planet Labs to launch two prototype satellites by early 2027.