A useful paper on the case for AI data centres in space [pdf] (starcloudinc.github.io)

A recent white paper from Starcloud proposes the establishment of orbital data centers to address the urgent energy demands of artificial intelligence development. As traditional data centers face rising energy costs and regulatory hurdles that limit their scalability, the paper argues that moving these computational resources into space could provide a sustainable, eco-friendly solution. With access to abundant solar energy and minimal atmospheric interference, space-based data centers could achieve energy generation costs as low as $0.002 per kWh, drastically reducing operational expenses compared to terrestrial facilities. This could enable the development of gigawatt-scale clusters essential for training next-generation AI models. The implications of this concept are profound for the AI/ML community. Currently, large AI models require massive computational power, which outstrips the capacity of existing energy infrastructures. By leveraging the unique advantages of space, including continuous solar energy availability and efficient cooling through passive radiative techniques, the proposed orbital centers could circumvent the logistical and environmental constraints found on Earth. As demand for computing power surges in the coming years, Starcloud's vision represents a strategic, innovative pathway to support the anticipated needs of AI, including artificial general intelligence (AGI), while addressing critical energy sustainability challenges.