jasdooit

Google has announced intentions to establish an artificial intelligence (AI) data center in space in order to utilize significant energy via solar power.

On the 4th, according to local time, Google unveiled its ‘Project Suncatcher,’ aiming to construct a data center that utilizes solar energy from space. With the increasing demand for AI, the energy consumption of data centers carrying out computations is also rising rapidly. Google mentioned, “The sun produces energy more than 100 trillion times the total human energy production,” and further noted, “In the future, space might serve as an ideal location for expanding AI computing.”

The fundamental strategy includes launching several small satellites fitted with extensive solar panels and Google’s proprietary AI chip, the Tensor Processing Unit (TPU), into low Earth orbit. Google stated, “Solar panels positioned in the correct orbit can generate energy up to eight times more effectively than on Earth.” These satellites will create an AI processing network via optical communication.

The main difficulty lies in enhancing the communication speed among satellites. This is due to the fact that hundreds to thousands of TPUs must share their computational outcomes in real time. Google has implemented a technique to boost communication speed by positioning the satellites very close to each other, with distances ranging from hundreds to thousands of meters. Google’s research team reported that they reached a bidirectional speed of 1.6 Tbps (1.6 terabits per second) during ground tests. For a space cluster, they aim to achieve at least 10 Tbps. Moreover, Google’s Trillium TPU functioned properly without significant harm even under the space radiation conditions equivalent to a 5-year mission.

Google revealed intentions to work with the private space firm Planet Labs in order to send two experimental satellites into orbit by the beginning of 2027.