27 months
Efficiency deflation
conversion in progress
The lagged path from AI capex to index earnings breadth Hyperscaler capex flows to compute orders, waits through an 18-36 month revenue lag, then books as earnings that decide index breadth; cheap AI mentions bypass the plumbing straight into chatter breadth. cheap signal bypasses the plumbing AI mention on call near-zero-cost · 65% chatter breadth chatter ~65% Hyperscaler capex $660–690bn guided · 2026 MSFT · GOOGL · AMZN · META ORCL · AWS · Google Cloud AI seller orders Nvidia ~$46.7bn (Q2'26) AI & cloud revenue 18–36 month lag supply-constrained ~$80bn backlog gated by power Index EPS breadth durable test · 12–24 mo 10% earnings breadth beyond AI sellers? capital deployed now vs revenue now $660–690bn capex guided for 2026 <$35bn pure-play AI revenue ≈1/20th OpenAI ~$20bn ARR · Anthropic >$9bn