Goldman and JPMorgan launch 4 AI debt baskets for institutional trading

Wall Street's two largest trading desks now let investors bet on AI infrastructure debt with a single trade.

Goldman Sachs Group Inc. and JPMorgan Chase & Co. launched products this week allowing investors to trade baskets of AI-related bonds in single transactions, as hyperscaler debt issuance for artificial intelligence investments accelerates.

The products allow hedge funds and other investors to manage sector-specific risks or express views on AI debt in a single swoop at one agreed price, according to a note from Goldman's trading desk reviewed by Bloomberg and people familiar with JPMorgan's offerings.

Goldman's basket includes 18 equal-weighted US high-yield issuers such as CoreWeave Inc., Applied Digital Corp., and Cipher Digital Inc., with pricing inquiries from $50 million to $250 million. The bonds carry an average yield of 7.45 percent and an average spread of 319 basis points, compared with 7.3 percent and 267 basis points for the broader high-yield market. JPMorgan launched three separate baskets Monday — one targeting investment-grade hyperscaler bonds from 11 issuers including Microsoft Corp., Meta Platforms Inc., and Amazon.com Inc., a second focused on 15 AI-related junk issuers, and a third covering 16 semiconductor and hardware companies including Nvidia Corp.

The new products come as fears over AI-linked spending intensify. Alphabet Inc.'s bonds came under pressure Thursday after the Google parent boosted its projected capital expenditures for the year and posted its first negative quarterly cash flow since going public more than two decades ago, highlighting the scale of the AI investment cycle driving demand for these hedging tools.

The AI Debt Financialization Play

Wall Street has rushed to create new hedging instruments as bondholder exposure to AI and data center projects climbs at an unprecedented pace. The structured baskets allow investors to take directional bets or hedge existing portfolios without buying individual bonds from each issuer — a process that would require multiple trades, separate pricing, and significant operational overhead.

Goldman's product can be executed as either physical bond purchases or total return swaps, giving investors flexibility in how they gain exposure. The average yield of 7.45 percent on the junk basket represents a premium of roughly 15 basis points over the broader high-yield market, reflecting the perceived risk of AI infrastructure companies that are burning cash to build data centers.

For JPMorgan, the three-basket structure segments the AI debt market by credit quality: investment-grade hyperscalers with strong balance sheets, high-yield AI infrastructure companies, and semiconductor firms that supply the hardware. This segmentation allows investors to target specific parts of the AI value chain.

What's at Stake for Bondholders

The products signal that Wall Street sees the hyperscaler debt buildup as a risk worth hedging against. Alphabet's $205 billion capital expenditure plan and its first negative free cash flow quarter since its 2004 IPO have rattled bondholders who worry that the AI arms race will pressure credit metrics across the sector. Microsoft, Meta, Amazon, and Oracle have all announced massive data center spending programs, with much of that investment funded through debt issuance.

If the AI investment cycle delivers returns as promised, these bonds will perform well and the hedging tools will have been unnecessary. If the spending overshoots demand, the ability to short AI debt through these baskets could prove valuable for institutional investors managing credit risk.

This article is for informational purposes only and does not constitute investment advice.