Goldman Sachs says AI debt is reshaping credit markets

Nearly $500 billion of AI-related debt had already been issued in 2026, Goldman Sachs Research estimated. For financing, syndicate, trading, research and client coverage teams, that means the work is moving from who will spend on AI to how the spending gets funded, priced and absorbed. How AI Debt Is Reshaping Credit Markets features Amanda Lynam and Zach Ablon, whose desks sit at the center of the new flow.

AI spending is becoming balance-sheet work

The basic economics are straightforward: data centers, chips, networking equipment, power infrastructure and long-dated compute capacity require enormous upfront capital. When that bill moves from equity hype to debt issuance, the credit market has to process more supply, more structure and more scrutiny. That changes the daily work inside Goldman because spreads, maturity profiles, covenant language and borrower quality all start to matter more than the headline narrative around artificial intelligence.

That volume is already large enough to alter deal flow across investment-grade and structured credit desks. Another Goldman Sachs estimate put the figure at $489 billion. For analysts and associates, that kind of pace usually means more live models, more issuer updates and more client calls that are about funding capacity instead of just technology demand.

The issuers are getting bigger, and the market is testing its limits

The debt wave is being driven most visibly by the largest technology companies. On June 29, Max Lukianchikov of Global Banking & Markets said the biggest tech firms had issued more than $170 billion in corporate debt in 2026, already more than they offered in 2025. He also said credit spreads remained very low, with investors still focused on yield rather than credit risk.

That is where the market starts to look less like a simple growth trade and more like a portfolio-construction problem. Goldman Sachs estimated Amazon, Google and Microsoft could add roughly $2 trillion in debt and still remain investment grade, but the U.S. bond market could comfortably digest only about $510 billion of that before investors pushed back hard. Goldman Sachs also estimated direct hyperscaler issuance could approach $400 billion in 2027, which would keep pressure on the public bond market even if the companies themselves remain highly rated.

For the desks that live in this market, the practical question is not whether the names are strong. It is how much supply can be absorbed without forcing concessions on price, tenor or structure. That is exactly the kind of issue that keeps syndicate teams, credit sales and trading desks busy when a single theme starts dominating issuance calendars.

What gets structured when AI capex turns into debt

Once AI spending is financed through debt, the structure of the borrowing matters as much as the borrower. Longer-dated bonds can fit the life of a data center better than short paper, and investors will care about how much of the stack sits at the operating company, how much is unsecured and how much cushion remains if buildout costs rise faster than cash generation. Those are not abstract questions for Goldman people who cover these names because they shape investor appetite, secondary-market liquidity and how much balance-sheet support is needed to get a deal done.

Alternative funding channels could include private credit and infrastructure funds. The public bond market will not be the only outlet if AI capex keeps rising. A borrower that might once have gone straight to a plain-vanilla bond deal may now weigh bank debt, public notes, private placements or fund capital depending on pricing and flexibility.

For junior bankers, that usually means more running between groups. Coverage bankers have to map the funding need, credit teams have to assess the leverage and cash flow story, and syndicate has to judge whether the market can clear the paper. If the cycle stays hot, it can mean more hours and more execution work, but also more chance to sit on the kinds of financings that build résumé value and exit opportunities into credit funds, private credit, structured finance or infrastructure investing.

Why Goldman is treating AI as a cross-business theme

The bank is not only a user of AI; it is treating AI buildout as a financing, trading and advisory theme that touches multiple businesses at once. Amanda Lynam, head of credit strategy research, and Zach Ablon, head of the credit sales desk in Global Banking & Markets, sit on two sides of that theme: one side is thinking about issuance capacity and investor behavior, while the other is watching how those securities actually move through the market.

If AI remains mostly an equity story, the work stays concentrated in a few advisory pockets. If it becomes a debt story, the opportunity spreads across financing, research, risk management, trading and asset management.