Cost to Insure Against AI Hyperscaler Defaults Climbs Sharply

The cost of credit default swaps tied to hyperscalers is climbing as more investors seek to hedge against a potential downturn in the AI boom, according to CNBC. Protection costs are rising fastest for companies carrying the largest debt loads, including cloud provider CoreWeave and Oracle.

For CoreWeave, the cost to insure $1 million of its bonds against default for five years has jumped from under $40,000 a year ago to $100,000 today, CNBC reported. CoreWeave's debt load has tripled over the past year. This week, its bond prices fell sharply, pushing yields above 15%. The company's CEO sent a memo to employees saying the market was acting "irrationally."

Oracle's debt load stands at $160 billion and is hovering just above junk-rated status. The cost of hedging against an Oracle default has risen 12% in a week, with a $1 million policy now costing $28,000. Part of investors' concern centers on a data-center project in New Mexico that has encountered setbacks. Oracle last month told a lender, including Blue Owl, that it would defer lease payments and other financial obligations tied to the deal if the datacenter does not come online as planned in 2028. It has already faced regulatory snags and local opposition, though Oracle still must pay some interim costs and rent for the space.

Blue Owl CEO Doug Ostrover addressed the hedging activity on a recent episode of the podcast Compound Interest. "There's a derivative market. If you want to make a bet that Microsoft could default and not pay their lease, you can go in the derivative market and do that," he said. "We don't. But it's available to us."

New debt deals are also feeling the strain. This week, a deal for Volta Infrastructure faltered as investors, already full up on data-center debt, demanded higher yields. Credit default swaps function as a form of insurance: buyers pay a premium and receive compensation if a default occurs.

The surge in hedging comes after record debt issuance over the past year, as companies raced to build the infrastructure and compute capacity needed for AI. That issuance has slowed in recent months amid rising interest rates and growing investor selectivity about whom they lend to.