AI Capex Tracker

Methodology

Data. Capex is capitalExpenditures from Alpha Vantage's quarterly CASH_FLOW statement, the same field (and the same sign convention) Free Cash Flow Yield uses. Revenue is totalRevenue from quarterly INCOME_STATEMENT, joined by exact fiscal-quarter-end date rather than re-swept here — it's read from the same 28-quarter (~7 year) history Margin & Leverage Cycle already keeps, so this page pays for one new full-index sweep (capex) instead of two. A company only counts once both series line up on at least 8 consecutive matched quarters.

The AI cohort. Six names: Microsoft, Alphabet, Amazon and Meta (the four hyperscalers whose combined capex guidance is itself a tracked market bellwether), Oracle (OCI's well-documented AI-driven capex ramp), and Micron (HBM memory fab capacity built specifically for AI accelerators). This is a judgment call, not a claim of completeness — chip designers with real AI revenue exposure but genuinely light capex of their own (Nvidia, Broadcom, AMD are largely fabless) are deliberately excluded, because their spending isn't where this page's signal shows up even though their revenue clearly is. A different, defensible list would give somewhat different numbers; the point is the shape of the trend, not this exact roster.

The structural break test. Aggregate quarterly capex intensity (sum of capex across the group ÷ sum of revenue, not an average of ratios) is regressed against a plain quarter index, split into two equal-length halves — oldest vs. most recent, matching Factor Analysis's equal-sample-size convention so neither half gets more statistical power than the other — and a Chow test asks whether the trend's slope genuinely differs between the two halves, the same test Factor Analysis runs on momentum persistence. A significant break in the AI cohort's slope, without a matching break in the rest of the index, would be real evidence of a distinct AI-driven capex cycle rather than a broad-based one. The opposite pattern, or no break in either, would argue against the "supercycle" framing.

The growth scatter. Pearson and Spearman correlation between trailing-12-month capex growth and trailing-12-month revenue growth, across every S&P 500 constituent with enough history — the same two-method check as every regression on this site. A positive, significant relationship suggests capex increases are broadly accompanying revenue growth; a weak or flat one suggests spending and monetization are, at least so far, only loosely connected across the index.

Sector and company name come from Sector Beeswarm's own weekly metadata. This page is a one-time snapshot as of the date below, not a recurring refresh, since cash flow and income statements only change when a company files a new 10-Q or 10-K.