Inside the AI 2040 Plan to Physically Restrict Frontier Compute
The AI 2040 proposal demands 1 Mbps air gaps, stripped data center cables, and joint US-China encryption keys to delay superintelligence.
Fruitfly Hard Takeoff, Parker's Vineyard, Personal Agents
With John Coogan, Jordi Hays, Thijs Simonian, Guy Oseary, Alex Heath, Mitesh Agarwal · Sunday, September 13, 2026
John Coogan and Jordi Hays discuss policy blueprints for pausing frontier AI training, the existential risk debate, and the simulation of fruit fly brains in software. Guests Thijs Simonian of OpenAI's robotics team, Guy Oseary and Alex Heath of Sound Ventures, and Mitesh Agarwal of Positron AI join to explore affordable robotics, early-stage venture investing, and custom memory-focused inference chips designed to compete in Nvidia's ecosystem.
Key takeaways
- The AI 2040 proposal aims to delay superintelligence from 2028 to 2040 by enforcing a mandatory pause on all frontier training runs. Read more →
- Guy Oseary backed both Anthropic and OpenAI before broad venture consensus formed, applying the same talent identification instincts he used to sign Alanis Morissette and Muse. Read more →
- Running Google's Veo video model requires four Nvidia H100s for a 10-second clip, exposing how video and long-context inference are bottlenecked by memory capacity and bandwidth rather than raw compute. Read more →
- Consumer robotics is repeating the desktop 3D printing curve, shifting from high-cost industrial cells to open-source hardware like the 3D-printed Hugging Face SO-100 arm. Read more →
4 articles from this episode
Guy Oseary on David Geffen's Racehorse Rule for Tech Bets
Guy Oseary and Alex Heath explain why Sound Ventures backed OpenAI early and how Geffen's racehorse blinders filter market noise.
Why Positron AI Bets Against Nvidia's HBM Monopoly
Positron AI COO Mitesh Agarwal explains why custom inference silicon must trade expensive HBM for commodity memory to win hyperscaler workloads.
OpenAI's Thijs Simonian on Low-Cost Robotics and Visual Planning
Thijs Simonian explains how OpenAI couples Codex with open-source arms like the SO-100 to bypass vision latency and token bottlenecks.
More TBPN episodes
- Washington Hones in on AI Safety, What Are AI Doomers Proposing | Diet TBPNSep 13 · 5 articles
- OpenAI Cracks Navier-Stokes, Tech Trailers, AI Jobs Apocalypse on Hold | Diet TBPNSep 13 · 4 articles
- Apple’s iPhone Duo, Meta’s Muse AI Agent, Anthropic Researcher Quits Over AI Doom | Diet TBPNSep 13 · 5 articles
- Some Personal Agent News, CharlestonAI Founder Joins, 10%, Mikey Shulman, Keith RaboisSep 13 · 7 articles