Hyperscalers Now Competing With US Treasury for Capital, Driving Up Government Borrowing Costs

A structural shift is happening in fixed income. Tech giants like Alphabet (NASDAQ:GOOGL | GOOGL Price Prediction) and Amazon (NASDAQ:AMZN) are issuing investment-grade debt at such scale that bond investors now treat them as sovereign-comparable borrowers. On a recent episode of Slate Money, the hosts talked about how hyperscalers are issuing so much debt that “they’re competing with treasuries now.”

The numbers are striking. The panel pointed to a trillion dollars of AI-related debt issued so far, mostly concentrated among hyperscalers rather than spread across speculative borrowers. Alphabet alone has tapped the senior unsecured note market repeatedly with $31.1 billion in Q1 2026, $24.8 billion in November 2025, and $12.5 billion in May 2025. Amazon’s long-term debt jumped to $119.1 billion from $65.6 billion year over year, with cash from financing activities of $52.767 billion in Q1 2026 alone.

The Yield Math That Reshapes Treasury Demand

Host Felix Salmon walked through the mechanics. If a conservative buyer can earn yield from a near-AAA mega-cap company like Amazon or Alphabet, the U.S. government must compete for the same dollar. As Salmon put it, “If investors can get, I don’t know, 6% from Alphabet, then why would you lend to the government at less than 5%?” He confirmed the Treasury-competition story is “absolutely true.” The current curve illustrates the gap. As of May 8, 2026, the 10-year Treasury yields 4.38%, and the 30-year sits at 4.95%.

Debt Funds AI CapEx

The borrowing funds an unprecedented infrastructure cycle. Alphabet guided to $175-$185 billion in 2026 capital expenditures, while Andy Jassy committed Amazon to “about $200 billion in capital expenditures across Amazon in 2026.” Alphabet’s Q1 capex of $35.674 billion more than doubled year over year, and free cash flow fell 46.63%. Sundar Pichai is pointing to a Google Cloud backlog of over $460 billion as the demand signal underwriting the spend.

The Bull and Bear Cases

Salmon’s defense rests on credit quality. He argued the buyers here are “relatively conservative,” and the paper is “incredibly safe” because issuance is concentrated among “the really deep-pocketed sort of ultra highly rated companies.” Elizabeth Spiers pushed back on Larry Fink’s dismissal of bubble risk, noting “so much of the data center buildout is built on debt” and warning against being “Wile E. Coyote running off the cliff and then just never looking down.”

If hyperscaler debt issuance continues to pull conservative capital away from Treasuries, the government’s marginal borrowing cost rises, which could drive higher interest rates. Higher rates ripple through almost everything tied to borrowing costs. Higher mortgage rates would make homes less affordable, and businesses would face higher financing costs, which can slow hiring and expansion. The irony is that AI is supposed to improve productivity and lower costs across the economy, but the infrastructure race behind it may initially have the opposite financial effect by increasing competition for capital and pushing rates higher.