Devoured - April 24, 2026
Oracle's Deluge of AI Debt Pushes Wall Street to the Limit (5 minute read)

Oracle's Deluge of AI Debt Pushes Wall Street to the Limit (5 minute read)

AI infrastructurefinance Read original

Oracle's $300 billion AI data center partnership with OpenAI has saturated Wall Street's debt markets, forcing banks to reject new projects and pushing developers to find alternative tenants or financing structures.

What: Oracle's massive deal with OpenAI to build hundreds of billions in data centers has overwhelmed banks' capacity to provide construction loans, as financial institutions hit regulatory limits on how much exposure they can have to a single counterparty, causing some developers to switch tenants (like Crusoe leasing to Microsoft instead) or move to bond financing instead.
Why it matters: The financing bottleneck reveals a critical constraint on AI infrastructure buildout beyond power grid issues and public opposition—even when demand exists, debt markets can only absorb so much risk from a single company, particularly one like Oracle with weaker financials than Google or Microsoft, potentially slowing the data center construction that AI companies desperately need to scale.
Takeaway: Developers planning large infrastructure projects should diversify tenant commitments and financing sources early to avoid hitting counterparty concentration limits that can block or significantly delay funding.
Deep dive
  • Banks like JPMorgan struggled for months to syndicate billions in construction loans for Oracle-leased data centers in Texas and Wisconsin, as institutional investors hit regulatory limits on single-counterparty exposure
  • The concentration problem forced at least one developer (Crusoe) to switch from Oracle to Microsoft as tenant for an Abilene, Texas expansion after lenders refused to finance more Oracle exposure
  • Oracle-related project finance deals are among the largest ever: $10 billion for Crusoe's original Abilene site, $38 billion for Vantage's Texas/Wisconsin campuses, and $18 billion for Stack's New Mexico facility
  • Oracle plans to raise $50 billion in stock and bonds for 2026 needs, but Morgan Stanley analysts estimate the company still requires over $100 billion more for 2027 and early 2028
  • Big tech companies are projected to spend $3 trillion on AI through 2028 but can only self-fund about half from cash generation, making debt access critical to AI infrastructure buildout
  • Oracle is in a comparatively weaker financial position than rivals like Google, Microsoft, and Meta—it has a lower investment-grade credit rating, more existing debt, and is currently burning cash
  • The cost of protecting Oracle bonds against default via credit-default swaps quadrupled between late September and late March 2026, though it has declined slightly since
  • Most of the borrowing was structured as short-term construction loans by data center developers with Oracle as tenant and OpenAI as subtenant, keeping the debt off Oracle's balance sheet
  • Vantage's Texas and Wisconsin loans took until Q4 2025 to largely syndicate and required more than 50 lenders to achieve successful distribution levels
  • Related Digital's Michigan data center campus chose Bank of America as lead arranger partly because it had less Oracle exposure than competing banks, and switched to bond issuance after seeing the construction-loan market struggles
  • Wall Street is generally providing flexible financing for the most creditworthy tech companies like Google, Microsoft, and Meta, but Oracle's financial profile makes lenders more cautious
  • Any slowdown in data center construction would hamper AI companies already hitting limits on what they can offer users as computing demand exceeds supply
Decoder
  • Counterparty exposure limits: Regulatory and internal risk rules capping how much money a bank or investor can lend to or have tied up with a single borrower or tenant
  • Syndication: The process where a lead bank distributes portions of a large loan to other lenders to spread risk and free up balance sheet capacity
  • Project finance: Loans structured around a specific project (like a data center) where the debt is secured by the project's assets and future cash flows rather than the developer's overall creditworthiness
  • Credit-default swaps (CDS): Insurance-like contracts that pay out if a company defaults on its bonds; rising CDS costs indicate markets see increased default risk
  • Investment-grade rating: A credit rating indicating relatively low default risk, typically BBB-/Baa3 or higher from rating agencies; Oracle has this but at a lower level than tech giants
  • Burning cash: Spending more cash than the company generates from operations, requiring external financing or asset sales to fund activities
Original article

Oracle's $300 billion megadeal with OpenAI is testing the limit of Wall Street's appetite for debt tied to the datacenter boom. Banks have struggled for months to spread the risk of the billions of dollars in loans they made to build data centers leased to Oracle in Texas and Wisconsin. Bank balance sheets are now clogged, constraining the financing prospects of future projects tied to Oracle and OpenAI. Silicon Valley needs access to debt to meet its goals for AI-related spending, but so far, Wall Street is largely giving a blank check for the AI ambitions of the most creditworthy tech companies.