Google And SpaceX Want To Move AI Data Centres Into Orbit
Google and SpaceX are reportedly discussing orbital AI data centres as tech firms search for new ways to power growing AI workloads.
Artificial intelligence is consuming so much electricity that Silicon Valley is now seriously considering moving parts of its computing infrastructure into space. Sources reveal that Google and SpaceX are in discussions around launching orbital data centres designed to handle future AI workloads outside Earth’s atmosphere.
The idea, once treated as science fiction, is increasingly being explored as technology companies confront growing limits around land, power availability, cooling, and infrastructure expansion for AI computing.
The discussions reportedly centre around Google’s “Project Suncatcher”, an initiative aimed at developing space-based computing systems powered by continuous solar energy. SpaceX would potentially provide launch infrastructure and orbital deployment capabilities as part of the effort.
Why AI companies are looking beyond Earth
The explosion of generative AI has transformed data centres into some of the world’s most energy-hungry industrial facilities. AI training and inference systems now require vast clusters of GPUs and accelerators running continuously, consuming enormous amounts of electricity while generating equally enormous amounts of heat.
That has triggered a race among technology companies to secure power, cooling, and infrastructure capacity. Google, Microsoft, OpenAI, Meta, Amazon, and Anthropic are all rapidly expanding AI infrastructure footprints, often competing for the same energy resources and semiconductor supply chains.
Orbital data centres are being explored as one possible long-term solution. In space, satellites could theoretically operate on nearly uninterrupted solar energy without relying on terrestrial electrical grids.
Reports suggest Google’s concept involves solar-powered satellites equipped with AI chips, including Google Tensor Processing Units, operating as a distributed computing infrastructure in orbit.
The logic behind orbital computing
Space-based AI infrastructure offers several theoretical advantages over conventional terrestrial data centres. The first is power generation. Satellites in orbit can access sunlight almost continuously, particularly in carefully selected orbital paths.
That creates the possibility of persistent solar-powered computing without depending on increasingly strained Earth-based grids. The second is land usage. Modern hyperscale data centres require massive physical campuses, extensive cooling systems, and increasingly difficult permitting processes. Orbital infrastructure bypasses those geographic limitations entirely.
The third is climate pressure. AI companies are facing rising criticism over the environmental footprint of large-scale compute infrastructure. Space-based systems could eventually help reduce pressure on terrestrial energy networks, particularly if renewable energy generation becomes constrained.
Companies increasingly view orbital infrastructure as a possible long-term answer to the AI industry’s growing compute bottleneck.
The engineering challenges remain enormous
Despite the excitement, orbital AI data centres remain highly experimental and economically uncertain.
Cooling is one of the biggest obstacles. Data centres generate immense amounts of heat, and cooling systems on Earth rely heavily on convection through air or liquid systems.
In space, heat dissipation becomes far more difficult because vacuum environments rely primarily on radiative cooling, which is significantly less efficient. Launch economics are another major hurdle. A recent analysis cited by TechCrunch estimated that a 1-gigawatt orbital data centre could cost roughly $42.4 billion, nearly three times more than a comparable Earth-based facility.
Academic research published in April 2026 similarly concluded that orbital data centres become economically viable only if launch and spacecraft construction costs fall dramatically below current levels.
There are also operational concerns involving satellite manufacturing, maintenance, radiation exposure for chips, communications latency, and the challenge of transmitting enormous amounts of data between orbit and Earth-based networks.
Google and SpaceX are already deeply linked
The reported talks also reflect the increasingly intertwined relationship between Google and SpaceX.
Google invested roughly $900 million into SpaceX in 2015 and reportedly still owns a stake of about 6.1 % in the company. Google executive Don Harrison also serves on SpaceX’s board.
SpaceX, meanwhile, has been aggressively positioning itself as a future AI infrastructure player rather than only a rocket company. Elon Musk has repeatedly discussed orbital computing as a potential long-term growth area for the business.
The company recently signed a major compute agreement with Anthropic and has reportedly filed plans involving massive satellite expansion to support future AI infrastructure ambitions.
The bigger AI infrastructure race
The discussions underline how rapidly AI infrastructure is becoming a geopolitical and industrial-scale competition. Technology companies are now racing not only to build better models, but also to secure the energy systems, semiconductors, launch infrastructure, and physical capacity required to sustain AI growth over the next decade.
Orbital computing remains speculative today, but the fact that companies like Google and SpaceX are actively exploring the idea signals how severe Earth’s infrastructure constraints are becoming in the AI era.
Whether orbital data centres become commercially viable or remain an ambitious experiment, one thing is already clear: the AI industry is beginning to think far beyond traditional data centres as it searches for the next frontier of compute.