Tech giants have borrowed roughly $500 billion since the start of 2026 to finance the chips, servers and data centers underpinning the artificial intelligence buildout, a debt binge that is increasingly rattling credit markets as investors question whether the returns will justify the borrowing.
Goldman Sachs expects the pace to accelerate further in 2027, projecting total AI-related tech borrowing could reach $1.2 trillion as Google, Meta, Amazon, Microsoft and others continue raising debt to fund infrastructure spending.
Credit default swap prices for companies including Microsoft, Nvidia, Meta and Amazon have surged since fall 2025, in some cases more than doubling, a signal that investors are growing more skeptical about whether the AI boom will generate returns large enough to justify the scale of borrowing behind it.
The skepticism has shown up directly in borrowing costs. Meta has had to offer yields above 7% annually on recent debt, while riskier cloud and data center specialists have had to pay more than 9% to attract lenders, a sharp increase from the financing terms available to these companies earlier in the AI spending cycle.
Global AI-related debt issuance plunged to $23 billion in September, down sharply from $113 billion in June, a steep drop that suggests either lenders are growing more cautious or the largest tech companies are pausing to reassess their financing plans amid the broader repricing of risk.
The combination of surging CDS prices, rising yields and a sudden pullback in issuance points to a credit market increasingly uneasy about how much debt the AI buildout can absorb before investors start demanding clearer evidence that the spending is translating into durable returns.
