Core Memory

He Comes Promising AI Freedom For All

59 snips
May 16, 2025
Guillaume Verdon, founder of Extropic and known as Beff Jezos, dives into revolutionary concepts of thermodynamic computing and its potential to transform AI. He discusses the significance of transitioning from traditional computing paradigms to probabilistic systems, while advocating for democratizing AI in a landscape dominated by corporate giants. Verdon also reflects on the evolution of open source ideals and critiques the power dynamics in AI, promoting competitive alternatives to ensure innovation and prevent monopolistic control.
Ask episode
AI Snips
Chapters
Transcript
Episode notes
INSIGHT

Thermodynamic Computing Explained

  • Guillaume Verdon explains thermodynamic computing creates new stochastic silicon primitives.
  • It aligns naturally with AI's probabilistic workloads, unlike traditional or quantum computers.
ANECDOTE

First Superconducting Probabilistic Chip

  • Guillaume showed their first prototype chip with three probabilistic bits made from superconductors.
  • It was a learning platform not commercially viable but crucial for pioneering this new computing paradigm.
INSIGHT

Scalable Silicon Probabilistic Chip

  • Their room temperature silicon chip achieves around 300 probabilistic bits using manufacturing at scale.
  • Hybridizing traditional AI with probabilistic hardware could cut GPU compute by 100x and boost energy efficiency by 1,000x or more.
Get the Snipd Podcast app to discover more snips from this episode
Get the app