causal
Inference shifting cheaper tokens increases compute
When AI natives and public companies route queries to cheaper open-source models, their token spend may fall, but total tokens and GPU hours rise because cheaper tokens let them consume more, so the routing shift is good for infrastructure demand.
causalpending
Speaker
GavinEvidence Quote
“The GPU compute hours probably are going up as you shift to these cheaper tokens you can use more of.”
Created: 11/12/2026, 9:48:03 AM
Your notes
Loading notes...