AI Data Centers Look to Space: A New Frontier for Computing
Major technology companies, including SpaceX's xAI and Google, are actively exploring the development of artificial intelligence (AI) data centers in outer space. This ambitious initiative seeks to alleviate the escalating demands for power, space, and water on Earth, while also circumventing growing public and political resistance to new terrestrial data center construction. Proponents, such as Elon Musk, envision nearly limitless renewable energy and natural cooling in orbit, though skeptics like OpenAI CEO Sam Altman voice concerns over current feasibility and cost-effectiveness.
The exploration of orbital AI data centers is driven by increasing infrastructure strains on Earth and public opposition to terrestrial construction.
The Rationale for Orbital Data Centers
The burgeoning energy and water consumption of AI data centers is placing significant pressure on Earth's infrastructure. Terrestrial facilities are massive electricity consumers; one analysis revealed that some areas near data centers experienced electricity cost increases of up to 267% over five years. Furthermore, large data centers can consume as much as 5 million gallons of water daily, a volume comparable to the usage of a town housing 10,000 to 50,000 people. These environmental and resource concerns, coupled with worries about pollution and local quality of life, have ignited political opposition and stalled approvals for new data center projects worldwide.
Proponents of space-based data centers highlight several compelling advantages:
- Power Access: Space offers superior access to solar energy. Solar panels in orbit could be up to eight times more productive than on Earth, capable of generating power almost continuously. Starcloud cofounder and CEO Philip Johnston emphasized the availability of nearly unlimited, low-cost renewable energy.
- Cooling: The natural vacuum and extreme temperatures of space can significantly aid in cooling technology.
- Land Use: Orbital data centers would completely eliminate the need for valuable land on Earth.
- Cost Efficiency: Elon Musk contends that transporting resource-intensive efforts to space is a logical solution due to available power and space. He has stated that scaling data centers would be "much cheaper" in space.
Company Initiatives and Regulatory Developments
A number of prominent technology companies are already pushing forward with space-based data center concepts:
SpaceX and xAI
Elon Musk's rocket company SpaceX and artificial intelligence firm xAI formally merged recently. SpaceX has filed plans with the Federal Communications Commission (FCC) for a million-satellite data center network. The FCC accepted this filing, initiating a public comment period. SpaceX announced its intent to launch such a constellation in February and has commenced hiring engineers for the project. Musk indicated that the xAI acquisition would accelerate the deployment of these orbital data centers.
Google
Google announced plans in November to test orbital AI data centers, with two test satellites slated for launch as early as next year. Google CEO Sundar Pichai reportedly suggested the company could begin deploying sun-powered data centers in space as early as 2027. Project Suncatcher, focused on implementing this technology, was reportedly unveiled in November 2025.
Starcloud
The Washington-based AI start-up Starcloud launched a test satellite equipped with an AI server aboard a SpaceX rocket in November. Philip Johnston, Starcloud's CEO, predicted that within 10 years, all new AI data centers will be in orbit, potentially mitigating public backlash against terrestrial construction.
OpenAI
Sam Altman, CEO of OpenAI, reportedly considered acquiring rocket company Stoke Space for orbiting data centers, signaling a broader industry interest in the concept.
Challenges and Differing Perspectives
While technical hurdles are acknowledged as engineering constraints rather than fundamental physics limitations, significant skepticism persists regarding the current viability and cost-effectiveness of large-scale orbital data centers.
Sam Altman's View
OpenAI CEO Sam Altman characterized the current concept of placing data centers in space as "ridiculous." He conceded that orbital data centers "could make sense someday" but cited high launch costs and the inherent difficulty of repairing computer chips in space as substantial obstacles. Altman stated that the technology is not yet viable at scale and will not significantly impact the industry within the current decade.
Operational Costs
In a podcast discussion with Elon Musk, Dwarkesh Patel questioned the assertion that space-based data centers would inherently be cheaper. He noted that power is not the sole operational cost, and other factors could significantly influence overall cost-effectiveness. Patel also raised critical concerns regarding the maintenance of GPUs in space should they fail during AI model training.
Economic Projections and Timelines
The global data center capacity is projected to reach 200 GW by 2030, representing approximately a trillion dollars in terrestrial infrastructure. Technology companies currently invest hundreds of billions of dollars annually in data centers. Projections for when space-based data centers might become economically competitive or widely deployed vary significantly:
- Elon Musk: Predicted that orbital data centers would become more cost-effective than Earth-bound ones within two to three years. He projected 2028 as a significant year for orbital data centers, asserting that within 30 to 36 months, space would become the most economically compelling location for AI infrastructure. Furthermore, he predicted that within five years, the annual launch and operation of AI capacity in space would surpass the cumulative total on Earth.
- Industry Experts: Deutsche Bank estimates that cost parity between terrestrial and orbital data centers will not be reached until well into the 2030s. David Bader, a distinguished professor of data science at the New Jersey Institute of Technology, suggested a more comfortable timeline of three to five years for the regular deployment of AI data centers in space, noting that the cost of launching satellites is decreasing while terrestrial AI data center costs are rising. Philip Johnston of Starcloud reiterated his prediction that all new AI data centers would be in orbit within 10 years.