ING Sells $500M Data Center Loan Portfolio as AI Debt Risks Intensify
nashnova research
ING is auctioning a $500 million portfolio of data-center loans with a $10 million minimum bid per tranche; the sale signals banks are accelerating the offload of AI-infrastructure credit risk to private-credit firms and insurers, driven by post-midterm fears of data-center construction curbs.
What exactly is ING selling?
The portfolio comprises 12 data-center loans totaling roughly $500 million, offered via a bids-wanted-in-competition (BWIC) process with a $10 million minimum per tranche and a November 19 deadline.
One $25 million loan is linked to the Stargate project — the New Mexico venture backed by OpenAI, SoftBank, and Oracle. The portfolio also includes exposures to operators Vantage Data Centers and CyrusOne.
This means → the package spans top-tier AI compute players down to mid-sized facility operators — not a piecemeal disposal, but a systematic clearing of data-center exposure.
Why sell now?
The Democrats' sweeping midterm victory raised expectations that some U.S. states may impose construction bans or stricter limits on data centers.
That fear triggered a sell-off on Thursday: the Nasdaq Composite fell 5.5% in a single session — the steepest one-day drop since April 2025 — and entered correction territory.
In plain terms = the election result cracked the default assumption that data centers are welcome everywhere, and banks responded in the most direct way possible — sell while the assets still have buyers.
How does the bank "originate-to-distribute" model work here?
SMBC strategist Hank Calenti explains: investment banks typically originate these loans, then distribute them to private-credit investors, asset managers, and insurers — buyers chasing higher expected returns.
In plain terms = the bank is a waystation, not the long-term risk holder. The institutions that buy in are the ones left holding the exposure.
Calenti adds that whether banks accelerate sales comes down to a trade-off: the mark-to-market loss on exit versus the balance-sheet space freed up for new lending. As he puts it: "This could be normal course of business, or it could be a lender looking to exit before the window shuts."
Is it just ING — or an industry-wide move?
Earlier this year, JPMorgan and Morgan Stanley were also reportedly exploring ways to reduce their data-center loan books.
Their tools include selling tranches to non-bank lenders through private channels and using significant risk transfer (SRT) instruments — structured deals that shift credit risk to investors and free up bank balance sheets.
This reflects not one bank's idiosyncratic call, but a collective repricing of AI-infrastructure debt across the banking sector. Whether ING's sale clears smoothly — and at what discount — will set a key reference point for how the market prices this class of risk.
市场有风险，内容仅供研究参考，不构成投资建议。