VCs debate AI moats: model lock-in via training loops vs harness and network effects
ruslansv · x · 2026-09-08
a16z partner @joshelman and @ruslansv debate AI moats: Josh argues that with multiple models at similar capability improving together, agent harness + context with network effects can be its own lock-in. Ruslan counters that model lock-in via training loops is far tighter than the commodity tech behind social networks, and only major labs can hope to achieve it.
Related event: VCs Debate AI Moats: Training-Coupled Harnesses vs. Generic Ones(4 posts)→
More from AGI Musings
- LEAP expert panel puts just 10% odds on AI cracking a Millennium Problem by 2027 — Afinetheorem · 2026-09-09
- LEAP panel of AI experts and superforecasters gives just 10% odds of AI solving a Millennium Problem by 2027 — Afinetheorem · 2026-09-09
- American Mathematical Society Weighs In: Navier-Stokes Progress Is a Milestone — AlexKontorovich · 2026-09-09
- Why 1k-10k agents per strong math solution may be the right order of magnitude — cephaloform · 2026-09-09
- Axios/Wired: OpenAI Began Navier-Stokes Push Sept 1; 10,000 Agents Delivered in 88 Hours — TheZachMueller · 2026-09-09
- OpenAI Says 10,000 Agents Cracked the Navier-Stokes Millennium Problem in 88 Hours — polynoamial · 2026-09-09