Google is preparing to send an artificial intelligence data center into space as the tech giant and a growing number of startups explore moving energy-intensive computing beyond Earth’s atmosphere.
The company’s Project Suncatcher is expected to take a major step toward that goal Oct. 1, when a prototype satellite developed by Google is slated to launch aboard a SpaceX Falcon 9 rocket from Vandenberg Space Force Base near Santa Barbara, California.
Stay informed.Stay ahead.
Join Washington Examiner for unlimited access to the news, analysis, and commentary that matter most.
Already a member? Log in
The satellite, dubbed the Minimum Viable Product, is an experimental precursor to the larger orbital data centers Google envisions. It will carry four specialized computer chips known as tensor processing units, or TPUs, with computing power roughly equivalent to that of a single data-center server.
The satellite will be designed to answer simple AI queries and is expected to operate for about a year, though it could remain in orbit for as long as six years before eventually reentering Earth’s atmosphere and burning up.
Google envisions eventually deploying compact constellations of solar-powered satellites equipped with its TPUs and connected through optical communications links. The company said the approach could provide a way to scale AI computing while reducing the demand for land and other resources needed by terrestrial data centers.
NASA Administrator Jared Isaacman has also said moving data centers into orbit could have “immense benefits,” particularly by allowing companies to take advantage of nearly continuous solar energy.
The idea is gaining attention as AI companies face soaring demand for computing power and the electricity, water, and land needed to support data centers on Earth.
A Government Accountability Office report released in April said space-based data centers could reduce the land, electricity, and water required by data centers on Earth. But the agency also identified major technological and economic hurdles, including the difficulty of dissipating heat in the vacuum of space and protecting computer hardware from radiation.
TRUMP’S AI STRATEGY WILL DEFEAT AMERICA BEFORE CHINA EVEN TRIES
Large orbital data centers would also require substantial solar arrays, communications systems, and cooling infrastructure. Manufacturing and launching the satellites could be expensive, while adding enough equipment to meet the power and computing demands of a large data center could increase a spacecraft’s size and launch weight.
Google is not alone in pursuing the concept. Space-based computing startup Starcloud filed an application with the Federal Communications Commission in February seeking authorization for a constellation of up to 88,000 satellites that would operate as a distributed data center in orbit.
