SpaceX and Google May Be Planning the Future of AI Computing in Orbit

Google is reportedly exploring partnerships with SpaceX and other aerospace companies to develop orbital AI data centers, according to a Wall Street Journal report, as the company looks at ways to support future AI infrastructure beyond Earth-based facilities. The discussions are tied in part to Google’s Project Suncatcher initiative, which was revealed last year and aims to place satellites carrying Google Tensor Processing Units into orbit beginning in 2027.

The idea of running AI compute infrastructure in space has circulated for years, largely because orbital systems could theoretically benefit from constant solar power generation. But launch costs and operational complexity have kept the concept largely theoretical. SpaceX has increasingly emerged as the most realistic launch partner because of its scale in commercial space operations and its aggressive focus on reusable rocket systems.

Neither Google nor SpaceX has publicly commented on the reported talks. Still, the report arrives as SpaceX continues expanding its ambitions around orbital computing infrastructure. The company recently partnered with Anthropic on plans that could involve “multiple gigawatts of orbital AI compute capacity,” and earlier this year filed an FCC application seeking approval for up to one million data center satellites.

The economics behind space-based AI infrastructure remain difficult. According to the report, Google’s Project Suncatcher estimates place the break-even point for orbital data centers near $200 per kilogram in launch costs. Current pricing remains significantly higher. SpaceX’s Falcon 9 launches are estimated to cost roughly $2,700 per kilogram internally, while customer launch pricing can reach about $7,000 per kilogram through standard rideshare programs.

Even so, SpaceX’s reusable rocket program continues pushing launch economics downward. One Falcon 9 rocket recently completed its 34th launch, and analysts believe repeated reuse could eventually reduce costs further by spreading production expenses across many flights. In that scenario, fuel, maintenance, and launch operations would become the primary ongoing costs.

The concept has long faced skepticism, including from OpenAI CEO Sam Altman, who previously questioned whether the economics of putting large-scale AI compute in orbit could ever work. But the rapid expansion of AI infrastructure demand, combined with SpaceX’s launch dominance, has revived interest in whether orbital compute could eventually become viable for certain workloads.

SpaceX currently leads the commercial launch market by a wide margin. The company has completed 165 launches in 2025 after conducting 134 launches in 2024, according to the report. It has also placed 14,844 payloads into orbit overall and is nearing the point where it will have launched as many satellites as all countries combined since the beginning of the space age.

The report also suggests the talks could further boost investor expectations surrounding SpaceX’s anticipated IPO, which is already expected to become one of the largest public offerings ever if it moves forward.

This analysis is based on reporting from Tom's Hardware.

Image courtesy of MarketWatch Website.

This article was generated with AI assistance and reviewed for accuracy and quality.