Banks Hit Risk Limits on AI Data Center Loans
Major banks now approach their internal risk thresholds due to enormous loans for AI data centers. Institutions like JPMorgan Chase, Morgan Stanley, and SMBC work to shift credit risks from these financings to other investors. The fast expansion of AI infrastructure across the United States puts pressure on the banking sector. Loan amounts for new data centers have swelled beyond what single banks can handle without exceeding concentration limits.
Scale of the Oracle Financing Deal
JPMorgan Chase, the biggest bank in the United States by assets, leads in many large-scale financings. Morgan Stanley focuses on investment banking and has deep ties to tech sector deals. MUFG, or Mitsubishi UFJ Financial Group, ranks as one of Japan's largest banks with global reach. These players now confront volumes that test their capacities.
Methods to Reduce Exposure
Banks pursue options like outright loan sales and significant risk transfers to curb their involvement. In risk transfers, loans remain on balance sheets, but default portions move to credit funds, insurers, or investors for a fee. Matthew Moniot from Man Group described the issue to the Financial Times. He noted banks soon feel overwhelmed by the sheer sizes involved. Man Group manages assets and invests in various strategies, including credit.
Frank Benhamou of Cheyne Capital views these transactions as more hazardous than standard risk transfers. Cheyne Capital operates as a hedge fund with alternative investments. Few operators dominate the space. Loans concentrate heavily. Construction efforts often overrun budgets or fail outright.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
Political Hurdles in Maine
Beyond finances, political forces create further obstacles. Maine's legislature approved LD 307, a moratorium on data centers of 20 megawatts or larger. This pause would last until November 1, 2027. The measure also called for a council to assess effects on power users, the electrical grid, environment, and economy.
Governor Janet Mills rejected the bill on April 24, 2026. She cited blockage of a $550 million initiative at the old Androscoggin paper mill site in Jay. Mills anticipates over 800 construction jobs, at least 100 ongoing roles, and higher tax income from the project. The veto held firm on April 29, 2026. Mills instead issued an executive order for a 15-member advisory group. This panel must provide guidance on managing big data centers in Maine by January 2027.
Data centers demand vast electricity, sparking debates on grid strain and sustainability. Such projects promise jobs and revenue but draw scrutiny over resource use. Banks factor these external risks into their lending decisions for AI builds.
The push for AI infrastructure continues despite these strains. Banks adapt through syndication and transfers to sustain funding flows. Political landscapes evolve, with states balancing growth against local concerns.