Take: Pre-training Data Was a High-Volume, Low-Margin Game
himanshustwts · x · 2026-08-29
A comment noted that the AI frontier went through a phase where pre-training data was a high-volume, low-margin game for data labs. Despite the economics, labs continued the work for the love of the game.
More from Companies & People
- Claire Vo to host pre-summit AI demo happy hour — lennysan · 2026-08-29
- Anthropic automates 95% of internal analytics queries with Claude — adnan_hashmi · 2026-08-29
- Union leader: Data centers offer a generational opportunity for blue-collar workers — pmddomingos · 2026-08-29
- Report: OpenAI Built a Hidden Task System for ChatGPT Inside Excel/PowerPoint — ryanmerket · 2026-08-29
- Product Compass tracks bug fixes across different AI models — PawelHuryn · 2026-08-29
- Shipping fast with AI feels like managing world-class engineers who can never give an ETA — jh3yy · 2026-08-29