Moonshots with Peter Diamandis Elon's $5 Trillion Bet, the End of Human Drivers, and Chamath's Market Warning | EP #242
637 snips
Mar 26, 2026 
Salim Ismail, OpenExO founder and exponential org thinker, joins Alexander Wissner-Gross, Reified founder and computer scientist, and Dave Blundin, Link Ventures GP and deep-tech investor. They dig into Elon’s terawatt chip-factory bet, robotaxis and eVTOLs reshaping cities, AI compressing market moats, orbital compute, hidden frontier models, and even cryonics and panspermia.
AI Snips
Chapters
Books
Transcript
Episode notes
Model Distillation Keeps Compressing Intelligence Downward
- Alex Wissner-Gross said repeated model distillation still feels almost magical because smaller systems keep preserving a surprising share of larger-model capability.
- He explained that big models generate synthetic data for student models, raising the possibility of eventually compressing superintelligence into tiny local devices.
The New AI Moat May Be Data Not Models
- The panel argued baseline models are commoditizing, so durable advantage is moving toward proprietary data and higher-level frameworks built above the model.
- Dave Blundin said rivals can copy parameters more easily than specialized training data in fields like chemistry, surgery, and materials science.
AI Designed Chips Expand Hardware Rather Than Replace It
- The group saw AI-designed chips as recursive self-improvement escaping software and starting to redesign the physical substrate of computation.
- Dave Blundin argued this will not eliminate hardware engineers; it lets them create many use-case-specific chips that may be 10 times more efficient.
