JPMorgan, Morgan Stanley, SMBC and MUFG are quietly attempting to sell off pieces of the AI-linked debt they underwrote over the past year, using significant risk transfer deals that let a bank keep a loan on paper while shifting its riskiest tranche to outside investors. The banks are bumping up against internal concentration limits — too much of their balance sheets now ride on one sector holding up. The clearest signal is a $38 billion construction debt package tied to Oracle data centers in Texas and Wisconsin. A group of banks including JPMorgan and MUFG has spent more than six months trying to distribute it, and some lenders are now willing to sell pieces to non-bank buyers at a discount just to move the paper. Matthew Moniot, co-head of credit risk sharing at Man Group, told the Financial Times the deal sizes are "out of scale to anything we've thought about, ever." That is a risk-pricing professional saying the numbers have left the model. Oracle is the weakest link in the chain. Its five-year credit default swaps climbed to around 151 basis points in early October, levels last seen in 2009. S&P cut Oracle's credit rating to BBB- in July — one notch above junk. 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. The money is flowing to data center sites in Texas, Wisconsin and New Mexico. Demand for AI-linked corporate paper is thinning across the board. Bonds from Alphabet, Amazon, Meta, Microsoft and Oracle now trade at a premium to similarly rated non-AI corporate debt. Order books for their offerings have shrunk from roughly five times the amount on sale in February to under two times by July. Meanwhile, the 10-year Treasury yield hit 5.304% this month — its highest since 2002 — after a $39 billion auction cleared at 5.30%, the highest auction yield since 2000. The Federal Reserve's own September meeting minutes named AI-related corporate borrowing as a contributor to that move. The structural architecture matters more than any single number. Oracle, Meta, xAI and CoreWeave have used special purpose vehicles to keep data center debt off their own balance sheets. But the risk did not disappear — it migrated to the banks that financed the SPVs. Now those banks are trying to push it further down the chain to hedge funds and institutional investors, who are themselves getting pickier. Capital is rotating toward data centers leased to investment-grade hyperscalers like Microsoft and Google, and away from weaker credits. The pricing gap tells the story. CoreWeave-leased paper prices at 7.00% to 9.875%, against 5.70% to 7.875% for hyperscaler-leased bonds. A smaller, less capitalized AI infrastructure company trying to raise debt next quarter is now competing for the same nervous pool of lenders as Oracle — except without Oracle's balance sheet or customer base. The AI boom runs on the assumption that credit stays open indefinitely at reasonable terms. Six months of a bank struggling to sell $38 billion in loans is the clearest evidence yet that assumption is being tested in real time. Banks do not sell down debt they are confident in. They sell down debt when concentration risk threatens their own stability. The question is no longer whether the AI buildout will produce returns — it is whether the credit pipeline can stay open long enough for those returns to arrive before the debt matures.
