Are data centers in space physically possible, or just another overhyped idea? In this episode, we speak with Philip Johnston, CEO of Starcloud, about the technical and economic case for putting AI infrastructure in orbit. The idea has gone viral in recent months, drawing strong criticism from science communicators like Scott Manley, Kyle Hill, and Hank Green, but rarely with detailed engagement on the underlying assumptions. We examine whether space-based data centers can compete with terrestrial infrastructure, and what constraints actually matter: energy generation, cooling, launch costs, and manufacturing at scale. Johnston walks through the core economic model behind Starcloud, including assumptions about SpaceX’s Starship, the cost of solar power in orbit, and why removing terrestrial constraints like land use, permitting, and energy storage could fundamentally change how compute is deployed. We discuss the physics of radiative cooling in space, the challenges of operating GPUs in a radiation environment, and how orbital systems compare to Earth-based data centers in terms of efficiency and cost structure. The conversation also explores broader questions around AI’s growing energy demands, the limits of terrestrial infrastructure, and whether shifting compute off-world is a niche solution or a long-term inevitability. Whether you’re interested in space technology, AI infrastructure, energy systems, or the economics of large-scale computing, this episode offers a detailed look at one of the most debated ideas in modern engineering, and a rare opportunity to hear its strongest arguments laid out in full. Follow us for more technical interviews with the world’s greatest scientists:Twitter: https://x.com/632nmPodcastInstagram: https://www.instagram.com/632nmpodcast?utm_source=ig_web_button_share_sheet&igsh=ZDNlZDc0MzIxNw==LinkedIn: https://www.linkedin.com/company/632nm/about/Substack: https://632nmpodcast.substack.com/ Follow our hosts!Mikhail Shalaginov: https://www.linkedin.com/in/mikhail-shalaginov/Michael Dubrovsky: https://x.com/MikeDubrovskyXinghui Yin: https://x.com/XinghuiYin Subscribe:Apple Podcasts: https://podcasts.apple.com/us/podcast/632nm/id1751170269Spotify: https://open.spotify.com/show/4aVH9vT5qp5UUUvQ6Uf6ORWebsite: https://www.632nm.com Timestamps:00:00 - Intro01:12 - What is Starcloud?02:44 - Why do data centers need to go to space?06:15 - Can’t we just build more solar panels on earth?11:10 - Economic analysis of Starcloud19:56 - How does Starcloud’s cooling work?28:26 - Training an LLM in space32:07 - Addressing critics on space Twitter34:23 - Is Starcloud overfunded?35:59 - Will demand for data centers keep going up?38:11 - GPU lifespan and disposal in space39:47 - Bus structures41:43 - Starcloud’s origin and founders49:29 - Fundraising, Competition, and Meeting Expectations53:29 - Satellite size and collisions56:29 - Manufacturing Bottlenecks1:00:20 - Starcloud 1 tests1:01:57 - Acceleration after YC1:03:43 - Testing on Earth1:05:06 - Motivations for Starcloud1:06:45 - Data centers on the Moon1:08:12 - Interacting with AI companies1:08:18 - What’s next for Starcloud?1:14:01 - Other uses for Starcloud satellites1:17:56 - Lunar hotels and space elevators1:24:28 - Complementary business ideas to Starcloud1:29:51 - Philip’s competitive twin1:32:18 - Philip and Mike’s thoughts on YC1:34:45 - Advice for young entrepreneurs #datacenter #aidatacenter #starlink #spacex #falcon9 #starcloud