Cost of insuring against default by AI hyperscalers hits record levels

The price tag for protecting yourself against an AI hyperscaler going belly up has never been higher. Credit default swaps on the debt of the biggest AI infrastructure spenders surged to record levels in late July 2026, a clear signal that the bond market is getting genuinely worried about the sustainability of the industry’s capital expenditure frenzy.

Oracle’s five-year CDS spread touched roughly 212 basis points, meaning it now costs $212,000 per year to insure $10 million of the company’s debt against default. For context, Oracle is one of five hyperscalers, alongside Alphabet, Amazon, Meta, and Microsoft, that have been borrowing at a pace that makes even seasoned credit investors uncomfortable.

The numbers behind the nervousness

The spending trajectory tells the story. From 2020 through 2024, these five companies issued an average of around $28 billion in US corporate bonds annually. In 2025, that figure exploded to $121 billion. Goldman Sachs projects cumulative capital expenditure across the group will reach $1.15 trillion through 2027.

Making matters worse, Alphabet reported its first quarter of negative free cash flow since its IPO in 2004 during Q1 2026. The culprit was unsurprising: AI spending.

Total off-balance-sheet lease commitments reported by the hyperscalers are nearing $1.65 trillion. These are financial obligations that exist but don’t appear in the most commonly scrutinized debt metrics, creating a somewhat distorted picture of how leveraged these companies actually are.

Wall Street builds new tools for a new risk

JPMorgan launched a customizable CDS basket product in early 2026, specifically designed to let investors hedge against the collective credit risk of the five major hyperscalers. Trading volumes in single-name CDS for these companies have increased sharply since the product’s introduction.

The monetization gap

At the heart of investor anxiety is a simple mismatch: spending is happening now, but the revenue these AI investments are supposed to generate remains largely theoretical for many use cases. Credit markets don’t trade on faith. They trade on cash flow coverage ratios, debt-to-EBITDA multiples, and the ability to service obligations in the near term. When Alphabet, arguably the most profitable advertising business ever built, tips into negative free cash flow territory, it forces a reassessment of what “safe” corporate credit actually looks like in 2026.

The situation bears watching not because a hyperscaler default is imminent, but because credit markets are historically better at pricing risk than equity markets. When the cost of insuring against default hits records for companies that have never come close to defaulting, it’s worth asking what the bond market sees that the stock market might be ignoring.

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