Key Takeaways
- • ING is reportedly seeking buyers for a portfolio of 12 data center loans worth around $500 million, according to Bloomberg, as banks push to distribute credit risk tied to the AI infrastructure build-out.
- • The auction includes a $25 million piece linked to the Stargate project in New Mexico, which involves OpenAI, SoftBank and Oracle, along with loan assets involving Vantage Data Centers and CyrusOne.
- • The sale is structured as a bids-wanted-in-competition (BWIC) process, with bids of $10 million or more due by Nov. 19.
- • The transaction comes amid rising concern over AI-related debt and possible U.S. data center construction moratoriums following the Democrats' midterm gains, with the Nasdaq Composite falling 5.5% in a single session.
Summary
ING is reportedly seeking buyers for a portfolio of 12 data center loans worth around $500 million, Bloomberg reported on Friday, citing people familiar with the matter. The auction includes a $25 million piece linked to the Stargate project in New Mexico, which involves OpenAI, SoftBank and Oracle, as well as other loan assets involving Vantage Data Centers and CyrusOne. The process is a bids-wanted-in-competition (BWIC) auction, with bids sized at $10 million or more due by Nov. 19. An ING spokesperson declined to comment.
The sale reflects a broader effort by banks to move data center loans off their balance sheets. Investment banks typically use a lend-to-distribute model for such loans, according to Hank Calenti, global markets strategist at SMBC, who said buyers can range from private credit investors and asset managers to insurers seeking higher expected returns. Earlier this year, Wall Street lenders including JPMorgan Chase and Morgan Stanley were reported to be exploring ways to reduce their exposure to data center construction lending. According to a Financial Times report on May 3, these approaches include private sales of debt to non-bank lenders and significant risk transfers, which shift part of a bank's credit risk to investors to free up balance sheet capacity.
The move comes as investor concern grows over the scale of debt financing the AI build-out and whether it will generate adequate returns. Those worries intensified after the Democrats' sweeping win in the U.S. midterms, which could bring moratoriums or tighter restrictions on data center construction in certain states. On Thursday, AI-related stocks sold off sharply, with the Nasdaq Composite closing down 5.5%, its worst day since April 2025, and entering correction territory. Pruksa Iamthongthong, head of APAC equities at Aberdeen Investments, said near-term volatility could continue as markets assess potential delays to U.S. data center construction and constraints on compute capacity. Calenti said banks' decisions to sell more exposure will depend on the discount received versus the benefit of freeing capital for further lending, adding that the sales could be business as usual or financiers looking to exit before conditions tighten.
The sale reflects a broader effort by banks to move data center loans off their balance sheets. Investment banks typically use a lend-to-distribute model for such loans, according to Hank Calenti, global markets strategist at SMBC, who said buyers can range from private credit investors and asset managers to insurers seeking higher expected returns. Earlier this year, Wall Street lenders including JPMorgan Chase and Morgan Stanley were reported to be exploring ways to reduce their exposure to data center construction lending. According to a Financial Times report on May 3, these approaches include private sales of debt to non-bank lenders and significant risk transfers, which shift part of a bank's credit risk to investors to free up balance sheet capacity.
The move comes as investor concern grows over the scale of debt financing the AI build-out and whether it will generate adequate returns. Those worries intensified after the Democrats' sweeping win in the U.S. midterms, which could bring moratoriums or tighter restrictions on data center construction in certain states. On Thursday, AI-related stocks sold off sharply, with the Nasdaq Composite closing down 5.5%, its worst day since April 2025, and entering correction territory. Pruksa Iamthongthong, head of APAC equities at Aberdeen Investments, said near-term volatility could continue as markets assess potential delays to U.S. data center construction and constraints on compute capacity. Calenti said banks' decisions to sell more exposure will depend on the discount received versus the benefit of freeing capital for further lending, adding that the sales could be business as usual or financiers looking to exit before conditions tighten.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.