Perplexity AI XAI's Radical Plan: Data Centers In Space
16 snips
Feb 12, 2026 They unpack plans to put AI data centers into orbit and why space might be cheaper for massive compute. Technical hurdles like radiation, thermal control, and satellite networking come up. The role of Starship and lunar manufacturing in lowering costs is explored. Competition from cloud giants and startups racing for orbital compute is discussed.
AI Snips
Chapters
Transcript
Episode notes
XAI Splits Into Four Teams
- Jaeden recounts XAI reorganizing into four teams: Grok, software generation, Imagine, and MacroHard.
- He explains MacroHard will focus on full task automation and corporate-scale modeling led by Toby Flynn.
AI Is Already Mainstream At Scale
- AI features on platforms are mainstream and driving massive usage of image and video generators.
- Jaeden cites internal metrics claiming tens of millions of videos per day and billions of images in a month.
Space Could Be Cheaper For Massive AI
- Elon Musk argues orbital data centers become cheapest due to constant solar energy and fewer land constraints.
- He claims a path to scale compute beyond terrestrial limits if launch costs fall enough.