The banks that raced to underwrite the AI data center boom are now quietly trying to hand the risk to someone else. According to Financial Times reporting cited in a widely shared analysis this week, JPMorgan, Morgan Stanley, SMBC and MUFG have been exploring what bankers call significant risk transfers — deals that let a bank keep a loan on its books while shifting the riskiest slice to outside investors.
The move comes roughly six months after many of the same institutions were competing to finance data center loans tied to Oracle, Meta and other companies driving the AI buildout. The shift suggests lenders are bumping up against internal limits on how much AI-linked exposure any single institution is willing to hold. For more context on this story, see our ongoing latest AI developments.
The $38 Billion Loan That Won't Move
The clearest sign of strain is a single deal. A group of banks including JPMorgan and MUFG has spent more than six months trying to distribute $38 billion in construction debt tied to a data center project in Texas and Wisconsin leased to Oracle, according to Bloomberg. Some lenders are now willing to sell pieces of it to non-bank buyers at a discount just to move it off their books.
Six months is a long time for debt that was never underwritten to be held forever. Matthew Moniot, co-head of credit risk sharing at Man Group, told the Financial Times that the deal sizes involved are "out of scale to anything we've thought about, ever." Coming from an executive whose job is pricing exactly this kind of risk-transfer trade, the comment signals that AI-linked financing has outgrown the models banks normally use to size and distribute credit risk.
What Is a Significant Risk Transfer?
Significant risk transfers, or SRTs, are a well-established tool that has historically been used mostly for corporate loan portfolios. The bank retains the loan on paper — keeping the client relationship and the interest income — while paying outside investors to absorb losses above a certain threshold. In effect, the bank buys insurance on its own balance sheet.
Banks generally do not sell down debt they are confident in. They sell it down when they are worried about concentration, which is what "approaching internal limits" means in plain language: too much of this particular risk in one place.
Oracle's Widening Credit Stress
Oracle sits at the center of the story. The cost of insuring the company's five-year debt through credit default swaps climbed to around 151 basis points in early October, levels last seen in 2009, according to market data cited by Bloomberg. S&P had already cut Oracle's credit rating to BBB- — a single notch above junk — in July as the company's AI spending outpaced its cash flow.
The company is reportedly targeting more than $40 billion in additional debt and equity this fiscal year, on top of the $18 billion in bonds it sold in a single day last September, to keep funding its Texas, Wisconsin and New Mexico sites. Oracle is also suing Wisconsin's Public Service Commission over a rule that could force it to post more than $7 billion in collateral for its Port Washington data center campus — a fight that helped push the rating cut.
A Cooling Bond Market for AI Borrowers
Demand for AI-linked debt has thinned as supply has surged. Bonds from Alphabet, Amazon, Meta, Microsoft and Oracle now trade at a premium to similarly rated non-AI corporate debt, and order books for their offerings shrank from roughly five times the amount on sale in February to under two times by July, according to Bloomberg. A bond that once sold itself now requires a bank to work to place it.
Higher Rates Raise the Stakes
The broader rate environment has sharpened the problem. The 10-year Treasury yield hit 5.304 percent this month, its highest since 2002, after a $39 billion auction cleared at 5.30 percent — the highest auction yield since 2000, per Bloomberg. The Federal Reserve's September meeting minutes named AI-related corporate borrowing as a contributor to the yield move, though Capital Economics argues that inflation and government deficits are the bigger drivers.
Whichever explanation holds, the effect on data center financing is the same: the risk-free rate that anchors every AI construction loan just became materially more expensive.
What It Signals for the AI Buildout
None of this means the AI infrastructure boom is over. But the mechanics matter: when the banks closest to a lending boom start hedging their own bets, they are pricing in the possibility that the revenue expected to service all this debt — Oracle's cloud commitments, Meta's AI products, Microsoft's Azure growth — may take longer to materialize than the loans take to come due.
The AI debt trade is becoming a two-sided market. In February, investors were fighting for allocations; by October, banks were discounting paper to move it. How that market prices AI infrastructure risk over the coming quarters may say more about the durability of the buildout than any benchmark score or model release.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
