AI & Concentration Self-Test
top-10 mega-cap & AI-capex cluster.
Methodology & data sources
What this is. A directional self-assessment, not investment advice. It translates an equity profile into an estimate of how much of the total portfolio is effectively a concentrated position in the largest mega-caps and the AI-capex cycle.
The concentration math. Implied exposure ≈ public-equity % × passive share × ~33% top-10 index weight, scaled by a home-region tilt factor (a US / developed tilt lifts the AI-capex share, a diversified or EM tilt lowers it). The ~33% figure is the modelled weight of the top-10 / Magnificent-Seven cluster in major cap-weighted global and US equity indices; replace it with a confirmed current index-fact-sheet figure on each refresh. The AI-capex sub-share of that cluster is a modelled assumption.
Drawdown sensitivity. The "AI-capex unwind" applies a −35% shock to the AI cluster plus a −12% spillover to the rest of equity, expressed as a one-year drawdown in portfolio-percentage points. Illustrative, not a forecast.
AI-policy stat. "69% of large asset owners have no formal AI policy" is drawn from the Mercer 2025 Large Asset Owner survey.
Cadence. The index-weight seed is refreshed quarterly from public index fact sheets via pipelines/refresh_ai-concentration-self-test.py.
© UniversalAssetOwners.com · For institutional discussion only · ~33% index weight and AI sub-share are modelled · Figures illustrative