AI Risk Has Become Impossible to Diversify Away
TREE NEWS reports: Artificial intelligence has quietly become the single largest correlated exposure across global markets, and the institutions that manage the world’s retirement savings are discovering there is nowhere to hide. New York City Retirement Systems Chief Investment Officer Monte Tarbox recently turned down a private equity fundraising offer specifically because the product was too heavily weighted toward AI holdings. “When that day comes, we don’t want to be left standing there with someone asking, ‘Why did you say yes to everything?'” he said — a sentiment now shared across nearly every major institutional allocator on Wall Street.
How Deep the Exposure Runs
The numbers are striking. Goldman Sachs estimates AI infrastructure companies — chipmakers and hyperscale cloud providers — now account for roughly 40% of total S&P 500 market capitalization. Apollo Global Management data shows AI-related issuers represent nearly half of this year’s investment-grade bond issuance and 87% of venture capital funding. Just three chipmakers make up more than a quarter of emerging-market benchmark indices. In private markets, private equity has become nearly synonymous with AI startups; in infrastructure, nearly every major project points to data centers.
Even US Treasuries can be folded into the AI trade narrative: surging electricity demand feeds inflation expectations, which in turn move yields. The same handful of companies now spans multiple asset classes. Alphabet carries a $4 trillion market cap, $130 billion in outstanding debt, and infrastructure connecting more than 30 data centers worldwide. Circular financing arrangements — companies holding stakes in one another — further interweave risk exposures.
Why Traditional Diversification Is Failing
Morgan Stanley Wealth Management CIO Lisa Shalett warns that AI is compounding the diversification challenges asset allocators already faced. With inflation proving sticky, the traditional 60/40 stock-bond portfolio has repeatedly broken down, and AI’s deep penetration is worsening the positive correlation between equities and credit. JPMorgan’s cross-asset strategy team built four models simulating institutional portfolios and found that all now show significant positive correlation to AI-related risk factors. In an Invesco survey of 90 sovereign wealth funds, more than half named market concentration as their top AI-related investment risk.
Measuring the Unmeasurable
The first challenge is simply defining and measuring AI exposure — there is no unified standard. The Los Angeles County Employees Retirement Association conducted a full review combining bottom-up analysis with MSCI’s top-down methodology, estimating 8% to 19% of its holdings are AI-related. Finland’s Elo Mutual Pension Insurance, managing €36 billion, is using AI tools to track AI sensitivity in its public holdings. Bloomberg Intelligence’s global AI stock index has delivered roughly 11 percentage points of annualized excess return over the past two years — meaning limiting AI exposure for diversification’s sake carries a real performance cost.
Key Takeaways for Investors
- Correlation risk is the new concentration risk. Holding different asset classes no longer guarantees diversification when the same AI-linked companies dominate equities, credit, private markets, and infrastructure.
- Measurement comes first. Institutions are adopting “Total Portfolio Approach” frameworks that break down asset-class silos — CalPERS has formally embraced this. Individual investors should similarly map their true AI exposure across accounts.
- Diversification has a cost. With AI driving outsized returns, underweighting it can hurt performance. The goal is not avoidance but deliberate, measured allocation.
- Risk profiles go stale fast. As NYCRS’s Tarbox put it, “We need not only to increase the resolution of the microscope but to accelerate the frequency of assessment.”
