JPMorgan: AI Concentration Risk Has Spread to the Bond Market
nashnova research
JPMorgan strategist Gabriela Santos warns that AI exposure is now piling up on both the equity and bond sides of portfolios — concentration risk in multi-asset books is being underestimated, as July's 21% semiconductor sell-off already demonstrated.
How did AI risk end up in bonds?
Investment-grade bond issuance has broken records for four straight months. Alphabet and other tech giants have even issued century bonds.
This means → if you own these companies' stocks, your bond portfolio may carry the same names — AI exposure on both sides, compounding risk rather than diversifying it.
Santos flags a specific concern: special-purpose vehicles (SPVs) — structures that package specific assets for financing — backed by data-center leases are growing increasingly complex and demand case-by-case credit analysis.
What did July's sell-off reveal?
The Philadelphia Semiconductor Index fell 21% in July, its worst month since 2008. South Korea's KOSPI dropped 22% over the same period; Samsung and SK Hynix together account for roughly half the index.
In plain terms = positions were too crowded; once selling started, the stampede followed — the scale of the drop shows markets were far more concentrated than anyone assumed.
Santos draws a clear lesson: position sizing, leverage control, and diversification across AI sub-sectors are all essential — none is optional.
How much will AI infrastructure actually cost?
Santos estimates total AI-build capital expenditure across public and private markets at roughly $5.5 trillion.
Goldman Sachs is even more aggressive: AI data-center spending could top $900 billion in 2026, with a 2027 forecast as high as $1.4 trillion.
This means → the sheer scale of funding is large enough to become a systemic weight in bond markets — this is no longer just an equity story.
So what should investors do now?
Santos is not bearish on AI. She calls this build-out "very unique" because the impact is already visible in corporate earnings.
But her core message is clear: even committed AI bulls should start diversifying now, not wait for the next sell-off.
The assets she names as differentiated hedges — U.S. Treasuries, gold, and core real estate — share the lowest correlation with AI themes and can cushion a portfolio when concentration risk materializes.
市场有风险，内容仅供研究参考，不构成投资建议。