AI has taken over the stock market. The bond market is next
Major technological revolutions have historically been accompanied by booms in corporate bond markets. In the 19th century, the expansion of railways transformed corporate finance, shifting it from small banks to deep national and international bond markets capable of funding massive infrastructure projects. Today, the rapid build-out of artificial-intelligence infrastructure is triggering a similar surge in corporate debt issuance.
Since the start of the year, AI leaders including Meta, Nvidia, Oracle, Amazon, Alphabet, and SpaceX have collectively raised well over $150bn through bond sales to finance AI-related investment. Many of these offerings are the largest debt issues in each company’s history. Morgan Stanley estimates that AI-related investment-grade bond issuance in America will reach $350bn–400bn this year, representing nearly one-fifth of all high-quality corporate bond issuance. A further $50bn could come from high-yield (junk) bonds financing AI projects.
The sheer scale of borrowing has prompted concern. The Bank for International Settlements has warned that many AI investments may ultimately fail to generate enough cash flow to repay the debt used to finance them. Bond investing is often described as the “negative art” of deciding what to avoid rather than what to buy, making the assessment of AI-related credit risk particularly challenging.
Paradoxically, however, the AI borrowing boom has also strengthened parts of the corporate bond market. America’s five hyperscale cloud providers—Alphabet, Amazon, Meta, Microsoft, and Oracle—added $228bn of debt in the six months to March, nearly five times any previous two-quarter increase. Yet these firms remain among the strongest borrowers in the world. Microsoft is considered an even safer borrower than the U.S. government, while Alphabet, Amazon, and Meta continue to generate enormous cash flows. Only Oracle carries a more modest credit profile, though even it remains investment grade.
Because such highly creditworthy companies dominate issuance, average corporate bond spreads have remained remarkably low—around 0.8 percentage points above U.S. Treasuries, close to the narrowest levels seen in a quarter-century. Investors have therefore shown little concern about the overall increase in borrowing.
At the same time, markets are becoming more selective. Investors increasingly distinguish between debt issued by established technology giants and bonds financing riskier AI infrastructure projects. For example, QTS Data Centers, owned by Blackstone and building a massive server campus for Microsoft, initially paid a spread only 1.1 percentage points above comparable Microsoft debt. That premium has since widened to 1.6 points, reflecting growing caution toward projects whose success depends on future AI demand.
Riskier financing is also expanding rapidly. Companies such as CoreWeave, Nebius, and Iren have issued billions of dollars in junk bonds and convertible bonds to fund AI infrastructure. Yet The Economist argues that this growth in speculative debt may actually represent healthy market adaptation. Financial crises have historically emerged not from assets already recognized as risky, but from supposedly “safe” assets whose risks were underestimated.
The historical comparison is sobering. Railroad debt defaults following the Panic of 1873 contributed to a prolonged global depression. While no comparable catastrophe appears imminent today, the AI revolution is evolving so rapidly that today’s winners may quickly become tomorrow’s losers—and vice versa. Determining which companies will still be successful enough to repay bonds maturing in ten, thirty, or even one hundred years, as in Alphabet’s case, will require bond investors to exercise exceptional judgment. In the age of AI, identifying which borrowers to avoid may prove more difficult than ever.
Source:
You might also like:
