Google is in early-stage discussions with SpaceX to deploy data centers in orbit to handle artificial intelligence workloads, according to a Wall Street Journal report. The concept, which has both excitement and skepticism, is part of Google’s Project Suncatcher, an initiative to test AI compute in space. The company plans to launch two prototype satellites by early 2027 with Planet to demonstrate the feasibility of solar-powered, orbital AI infrastructure.
The talks are not exclusive; Google is also speaking with multiple rocket-launch providers. Meanwhile, SpaceX is preparing for a massive initial public offering later this year, targeting a valuation of about $1.75 trillion. CEO Elon Musk has been pushing aggressively to become the first to build space-based computing infrastructure, emphasizing advantages like near‑free cooling, consistent solar energy, and avoidance of terrestrial permitting and community opposition. Last week, AI company Anthropic agreed to use SpaceX’s Colossus 1 computing facility in Memphis and expressed interest in collaborating on multiple gigawatts of orbital data center capacity.
Despite the momentum, significant hurdles remain. A BitcoinWorld analysis notes that today’s terrestrial data centers are substantially cheaper once launch and satellite construction costs are factored in—adding a premium of several hundred percent per kilowatt of compute capacity. Technical challenges include radiation hardening, orbital debris risks, and the difficulty of performing maintenance in space. However, if launch costs continue to drop, especially with vehicles like SpaceX’s Starship, the economics could narrow over the next decade.
Google previously invested $900 million in SpaceX in 2015, giving the search giant a longstanding strategic interest. Neither company has publicly commented on the latest discussions, which sources describe as still in early stages. The conversations underscore a growing belief that space could eventually provide a scalable, sustainable location for AI compute as ground-based data center construction faces increasing land, power, and water constraints.
