Ex-AI Exec: Missing a Training Cycle Causes Temporary Gap; Labs Bet on High-Variance Moonshots
brianryhuang · x · 2026-08-09
Addressing concerns about AI labs' progress, former executive Aman Sanger acknowledges that missing a training cycle can create a surprisingly large temporary gap. However, he notes that outsiders often only see shipped models and a few prominent figures, mistakenly inferring they represent the entire lab.
In reality, large labs have many teams working on less visible foundational tasks that may only bear fruit one or two generations later. High-profile staff are often pursuing high-variance "moonshots" like new architectures, scaling approaches, multimodality, and agents. By definition, these projects have a high failure rate, so going a cycle without a competitive release is expected.
More from Companies & People
- AI Weekly: DeepSeek Infinite Loop, Google Exec Shakeup, and Model Sandbox Escapes — APPSO · 2026-08-09
- The Hardest Startup Challenge Today: Attracting and Retaining Great Talent — annbordetsky · 2026-08-09
- 68% of Employees Feed Sensitive Data to AI, Exposing Major Governance Gaps — shensi · 2026-08-09
- Satya Nadella: LinkedIn Merged PM, Designer, and Engineering Roles into Full-Stack Builders — aakashgupta · 2026-08-09
- DeepMind and Google Brain Hemorrhage Key Talent to OpenAI and Anthropic — tengyanAI · 2026-08-09
- ByteDance Reportedly Training Massive AI Model to Rival Top US Labs — pstAsiatech · 2026-08-09