20VC: Is Anthropic's Coding Business Worth $2 Trillion? | Should American Enterprises Work With Open-Source Chinese Models? | Why 80–90% of Neo-Labs Die in the Next 18 Months? with Eno Reyes, Co-Founder @ Factory
Summary
Harry Stebbings interviews Factory CTO Eno Reyes on AI’s value stack, arguing that smartest models often yield the cheapest outcomes. They discuss the rise of specialized, open-source models, the critical need for sovereign intelligence to avoid vendor lock-in, and why most niche AI labs will fail. Reyes emphasizes that businesses must own their data and workflows to thrive in this transformative era.
Topics discussed
Intro: The cheapest model vs. cheapest system and sponsors
Guest background and the economics of AI outcomes
Democratizing intelligence and the role of verification
Defining quality, management strategies, and future business models
Margin profiles, profitability, and the risks of model locking
Revising predictions and the shift to local open models
Commoditization of routing and the infrastructure layer
Context windows, continuous learning, and sovereign intelligence
Risks of outsourcing intelligence and the Cursor/SpaceX acquisition
Survival of neo-labs, sectoral lags, and model censorship
Open models vs. frontier labs and enterprise AI strategy
Valuation bubbles, hiring for rule-breakers, and team culture
Measuring talent value and avoiding burnout cultures
Enterprise sales as discovery and rejecting self-service subsidies
Quick fire: Nvidia, Meta, Linear, and competitor threat levels
Future of software generation and closing remarks
Listen ad-free on Castria