Investors dump longer-dated AI debt as Big Tech’s $159 billion borrowing binge tests market appetite

Big Tech has a new favorite ATM, and it’s the corporate bond market. Alphabet, Amazon, Meta, Microsoft, and Oracle have collectively raised $159 billion in corporate bond offerings through May 2026, blowing past the roughly $150 billion those same companies borrowed across the entire five-year stretch from 2020 to 2024.

The problem: investors are increasingly saying “sure, we’ll lend you money, but not for 30 years.” Longer-dated AI debt is getting sold off as buyers rotate into shorter maturities, spooked by elevated Treasury yields and the uncomfortable reality that nobody knows when these massive AI bets will actually generate returns.

The numbers tell a story of caution

Look at Amazon’s recent $25 billion bond sale. In normal times, a company with Amazon’s balance sheet would have investors tripping over each other to get in. Instead, the offering attracted orders of roughly 1.6 times the amount on offer.

The backdrop makes the caution understandable. Ten-year Treasury yields have been hovering around 4.49% to 4.5%, levels not seen in multiple years. When the risk-free rate is that high, locking up capital in 20 or 30-year corporate bonds requires a compelling story about future cash flows.

The 10-year to 2-year Treasury spread has flattened to approximately 0.29%, a signal that bond investors are growing uneasy about the long-term outlook.

Alphabet’s debt load tells the real story

Perhaps no single company illustrates the scale of this borrowing spree better than Alphabet. The Google parent’s long-term debt ballooned from $10.9 billion at the end of 2024 to $46.5 billion one year later, a more than fourfold increase in twelve months, driven almost entirely by the need to finance AI infrastructure.

Overall AI-related capital expenditures across the industry are projected at around $750 billion for 2026. Morgan Stanley has projected that global issuance of AI-linked debt could reach as high as $570 billion in 2026 alone.

The money is flowing into data centers, graphics processing units, and the sprawling physical infrastructure required to train and deploy AI models at scale.

Where crypto meets the AI debt wave

The intersection of AI infrastructure financing and digital assets is producing some unexpected crossovers. Core Scientific, once primarily known as a cryptocurrency mining operation, is seeking to raise $3.3 billion through a junk bond offering to fund its pivot from crypto mining to AI data centers.

The Core Scientific deal also highlights a different corner of the AI debt market. While Alphabet and Amazon are issuing investment-grade bonds, companies further down the credit spectrum are tapping high-yield markets to fund similar infrastructure plays.

What this means for investors

The selloff in longer-dated AI debt creates a two-speed market worth watching. Short-duration bonds from investment-grade tech issuers remain in healthy demand, as investors are willing to fund AI buildouts over three-to-five year horizons. The skepticism concentrates at the long end, where 20 and 30-year maturities force buyers to make assumptions about technology cycles that are inherently unpredictable.

For equity investors in Big Tech, the borrowing binge introduces a variable that hasn’t been part of the thesis for most of the past decade. Companies that were essentially debt-free are now carrying significant leverage. If AI revenue growth disappoints or takes longer to materialize than expected, those debt service costs don’t disappear.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.