Sep 16, 2026 · 9 min
Big Tech’s AI borrowing tests the bond market
How the bond market is handling AI risks
The debt financing the data-center boom could expose investors to risks that outlast today’s enthusiasm for artificial intelligence.
- 1Hyperscalers have issued more than $130 billion in bonds so far in 2026 to finance data-center expansion.
- 2Strong demand is beginning to give way to higher yields as investors weigh an unusually large supply of corporate debt.
- 3Bondholders face construction, leasing, interest-rate, regulatory, demand, and obsolescence risks before projects repay their debt.
Don't miss
The episode’s clearest warning comes when repayment risks extend beyond borrowing costs to leases, demand, regulation, and facilities becoming obsolete.
The brief
Big technology companies are borrowing heavily to build AI data centers, turning the infrastructure boom into a test of whether investors can absorb a flood of new corporate bonds.
The five major hyperscalers issued more than $90 billion in bonds over the previous year and more than $130 billion so far in 2026, competing with U.S. Treasuries for investor money.
Alphabet’s heavily oversubscribed bond sale shows demand remains strong, but investors are beginning to require higher yields and better compensation for taking on the supply.
Zachary Griffiths and Jonathan Mondillo explain how construction delays, lease problems, changing AI demand, regulation, interest rates, and technological obsolescence could threaten repayment.
The central tension is stark: investors keep buying the debt even as the long-lived facilities it finances could become uneconomic before their bonds mature.
Alphabet Inc.
Microsoft Corporation
Amazon.com, Inc.
Mean GirlsListen to the full episode and explore every guest, topic, and moment on PodLume.