causal

Test-time compute aligns spending with revenue

Shifting from pre-training to inference-time compute better aligns expenditure with revenue, because capex is spent in step with actual model usage rather than front-loaded into a 9 to 12 month training run whose revenue arrives later.

causalpending

Speaker

Modest

Evidence Quote

“this is much, much better for the hyperscalers.”

Source

Capital, Compute & AI Scaling— Invest Like the Best
Created: 11/18/2026, 6:42:24 AM

Your notes

Loading notes...