The recent surge in U.S. bond yields to levels not seen in two decades has sparked debate over its causes and consequences. While inflation remains above the Federal Reserve's 2% target, the yield increase since late August is not tied to rising inflation expectations. Instead, it reflects a climb in real yields, driven by strong economic fundamentals and unprecedented capital spending.
Fed Chair Kevin Warsh, in his September press conference, pointed to three factors: robust economic growth, a surge in capital expenditure, and geopolitical tensions. The first two are directly linked to the massive buildout of artificial intelligence infrastructure. This AI boom is now supplementing consumer spending as a key engine of U.S. growth, with the five major hyper-scalers—Amazon, Alphabet, Meta, Microsoft, and Oracle—expected to invest over $800 billion this year and $1.1 trillion next year.
Columbia Business School's Stijn Van Nieuwerburgh places this spending in historical context, estimating that planned investments could total more than $10 trillion from 2025 to 2032, or 3.6% of GDP annually. That would eclipse past infrastructure projects like canals, railroads, and highways. However, these figures may overstate the direct economic impact, as a significant portion will be spent on imports.
So far, the growth effect is largely from business investment, but as AI adoption spreads, productivity gains could become more entrenched. A Federal Reserve study shows a mixed picture: while some sectors see productivity gains, labor market impacts remain limited, consistent with a buildout phase rather than widespread disruption.
The scale of hyper-scaler spending is now so large that internal cash flows can't cover it. Morgan Stanley estimates that over half of the $3 trillion needed between 2025 and 2028 will come from external sources, including debt and third-party equity. Vontobel Asset Management reports that U.S. dollar bond issuance by these firms could hit $250 billion this year, with $400 billion globally, and over 40% of that debt has maturities beyond 15 years, reflecting the long life of data center assets.
This private-sector financing shift coincides with a federal budget deficit near 6% of GDP and record public debt, intensifying competition for capital. Construction outside data centers is already feeling the squeeze, and the housing market is vulnerable as mortgage rates climb above 7%. The key question is whether rising yields will dampen growth or coexist with it.
Current data suggest the AI boom is powering the economy, with manufacturing and services strengthening and unemployment low. A Financial Times survey of bond investors found most believe soaring yields are "not even close" to cooling the economy. My view is that yields are returning to normal after years of subpar growth post-2008, akin to the telecom-driven expansion of the late 1990s.
The Fed must now decide if the economy's potential growth rate exceeds its 1.8% estimate, which would justify higher long-term real yields, or if the capital competition is temporary. It also faces financial risks from the AI buildout's reliance on external funding, with the biggest danger being overly optimistic return expectations.
