About 34% of global fund managers now cite AI hyperscaler spending as the most likely source of a future systemic credit event
TIMOTHY A. CLARY / Getty Images
Concern over AI data center debt is rising on Wall Street's list of potential credit threats, according to Bank of America $BAC's latest global fund manager survey, with a growing share of investors worried that the pace of financing could trigger the next market shock.
Concern about AI hyperscaler capital spending has surged among investors: In Bank of America's latest survey, conducted earlier this month, that category was named the most likely source of a future systemic credit event by about 34% of fund managers — twice the proportion recorded in April. At 42% of respondents, U.S. private credit held its position as the leading worry, even as that share fell from 57% the month before. The survey polled more than 150 participants between May 8 and May 14.
Since the start of last year, U.S. investors have channeled more than $300 billion in borrowing to tech companies financing AI spending, according to Bloomberg. Bankers anticipate that figure will keep climbing. The underlying anxiety centers on whether companies can ultimately justify borrowing at this scale for infrastructure whose revenue potential has yet to be proven.
Speaking on Bloomberg TV, De Boltz — JPMorgan $JPM Chase's managing director of leveraged finance — described deal activity as having taken on a compounding momentum that is reshaping how lenders manage their balance sheets. "Everyone is working out where they're going to put that cash, and how much cash do they have to hold for these deals," De Boltz said. He added that lenders have grown more cautious about financing software companies that could be disrupted by AI, directing more capital instead toward businesses directly tied to the technology. "All of that cash is going into AI," he said.
The shift in sentiment comes as AI data center financing deals have grown larger and more complex, with debt structures that can route exposure to ordinary investors through widely held bond funds in retirement accounts. One deal currently in negotiation involves a $16 billion data center campus in Michigan, with about $14 billion of that figure structured as debt — illustrating the scale at which individual projects are being financed.
The remaining respondents spread their concerns across other categories: crypto and stablecoins drew 2%, Japanese government debt 4%, and U.S. consumer credit 6% as the most likely flashpoint for a credit event.
When asked about broader tail risks facing markets, inflation's potential return topped the list at 40%, followed by geopolitical conflict at 20%, a disorderly bond-yield spike at 18%, an AI bubble at 11%, and private credit at 6%.
The debt concerns are compounding other pressures on the AI data center industry. Public opposition to data center construction is intensifying across the U.S., driven in part by concerns over electricity costs. Community resistance stalled or halted 48 projects representing more than $156 billion in planned construction last year alone, according to Data Center Watch data cited by Fortune.
Join 500,000+ readers who start their day with Quartz.
By subscribing, you agree to our Terms of Service and Privacy Policy.
