Big Tech wants to build the future of artificial intelligence. Increasingly, it wants to do that with borrowed money.
A Bloomberg opinion piece argues that the corporate bond market already recognizes the risks of AI’s enormous capital spending. It also argues that the Federal Reserve should tread carefully so a market wobble does not turn into a self-reinforcing downward spiral.
The borrowing binge, by the numbers
In 2025, five hyperscalers (Google, Amazon, Meta, Microsoft and Oracle) issued approximately $108 billion to $121 billion in corporate bonds tied to AI-related investment, according to the research findings. Their annual average in prior years sat somewhere between $20 billion and $28 billion.
The IMF’s October 2026 report indicated that hyperscalers now account for 10% of global investment-grade bond issuance. Just two years earlier, that figure was less than 1%.
AI-linked investment-grade bonds from these companies carry weighted-average maturities of 16 to 17 years. The broader market average is approximately 10 years.
Why bond investors are getting picky
According to the research findings, investors are demanding higher yields on this debt because of uncertainty about when, and how fully, AI products will be monetized. Longer maturities amplify that uncertainty, since every extra year is another year for the business case to change.
The IMF’s October 2026 report warned that rising AI investment might not deliver the gains companies expect. It also suggested AI capabilities may not require as much spending as anticipated, which could undercut the returns on all this new infrastructure.
What the Fed is watching
In August 2026, Federal Reserve officials voiced concerns about the financing structures behind the AI buildout, with some describing them as potentially “too big to fail.”
Economists at the Kansas City Fed took a closer look in October 2026. Their analysis highlighted structural risks tied to concentrated exposures and to novel, high-leverage financing vehicles connected to AI.
The research findings also note that lending rates show signs of potential increases in response to this volatility.
What this means
For bond investors, with hyperscalers now making up 10% of global investment-grade issuance, a sentiment shift around AI could ripple through portfolios that hold broad investment-grade exposure. The research findings warn that the pricing of AI-related bonds is likely to stay volatile as long as monetization timelines remain unclear.
The IMF and Fed officials have both urged vigilance to avoid leverage making any downturn worse.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.