Compressing Machine-Building Labor Into Tokens
jwt0625 · x · 2026-07-18
The author proposes a hypothesis: if all engineering cognitive labor—from raw materials and rack design to mass deployment—could be represented and understood by intelligent systems in the form of tokens, the entire process of building machine intelligence might be unified into a computable scale.
They provide a rough estimate: this equates to about 50 million to 500 million hours of engineering effort, or 1 trillion to 30 trillion tokens. Based on the throughput of a GB300 rack, these tokens could be processed in 10 days to 1 year. The conclusion is that the true bottleneck isn't the compute power itself, but rather real-world representation and execution.
More from AGI Musings
- A multipolar AI race will not automatically make AI go well, repost argues — JeffLadish · 2026-07-22
- Humanoid robot sorting packages in a warehouse sparks debate over job loss — MonaJalal_ · 2026-07-22
- You can outsource thinking, but not understanding, in the age of agents — Yuchenj_UW · 2026-07-22
- India’s multilingual LLM edge, once obvious, is gone, the post argues — kmeanskaran · 2026-07-22
- AI media may be cleaned up with provenance tracking, notes, and prediction markets — NathanpmYoung · 2026-07-22
- Ryan Greenblatt says economists underestimate AI’s growth impact even in a 100 million worker scenario — RyanGreenblatt · 2026-07-22