Google is developing the Suncatcher project to power artificial intelligence data centers in orbit using solar energy. An experimental satellite named MVP, built and tested at Planet Labs in San Francisco, is scheduled to launch on October 1 from Vandenberg Space Force Base in California aboard a SpaceX Falcon 9 rocket. Google Vice President James Manyika stated that the refrigerator-sized satellite will pack sufficient computing power to handle basic AI queries directly from space.
Despite these initial steps, significant technological and economic hurdles remain for orbital infrastructure. Companies like SpaceX and Blue Origin are also exploring space-based data centers, but scaling remains a major challenge. Google manager Travis Beals estimates that approximately 10,000 satellites would be required to match the capacity of a single one-gigawatt data center on Earth. Additional obstacles include the lack of air for cooling, potential chip damage from cosmic radiation, and the necessity for launch costs to fall to roughly $200 per kilogram. Amazon founder Jeff Bezos has noted that it could take up to two decades for orbital data centers to become more cost-effective than ground-based alternatives.
This initiative matters because it highlights the growing energy and infrastructure constraints of terrestrial artificial intelligence and explores radical long-term solutions for powering future computing demands.