Dutch bank ING is putting roughly $500 million of data center loans up for sale in a single package, Bloomberg News reported Thursday, citing people familiar with the matter — the latest sign that nerves over AI-related lending are spreading through the financial system.

The portfolio includes a dozen loans being marketed through a so-called bids-wanted-in-competition process, according to the Bloomberg report. Reuters could not immediately verify the report, and ING did not immediately respond to a request for comment outside business hours. For more context on this story, see our ongoing artificial intelligence updates.

A Stargate Loan Is in the Package

The most eye-catching asset in the sale is a $25 million loan tied to a large New Mexico data center that forms part of the Stargate artificial intelligence infrastructure venture backed by OpenAI, SoftBank Group and Oracle, according to the report.

That detail gives the sale symbolic weight. Stargate is the flagship of the AI buildout — a multibillion-dollar program to construct the computing capacity that frontier models are expected to require — and a bank quietly shopping even a modest slice of its exposure to the project captures how cautiously some lenders now view the sector.

The rest of the portfolio includes debt linked to established data center operators such as Vantage Data Centers and CyrusOne, Bloomberg reported. Bids are due on November 19, with a minimum ticket size of $10 million.

What Data Center Debt Actually Finances

Data center loans sit at the base of the AI supply chain. They pay for the shells, power infrastructure and cooling systems of the buildings that hyperscalers and AI labs then fill with GPUs. Lenders have treated the sector as attractive because the facilities are typically backed by long-term leases from investment-grade tenants, which makes the credit look bond-like.

That logic has driven enormous issuance. As AI companies have raced to secure capacity, banks, private credit funds and institutional investors have financed construction at a pace the industry has never approached before. A Nikkei analysis published earlier this week estimated that hidden AI-related debt at major US tech companies ballooned 90% to roughly $3.2 trillion — a figure that includes the off-balance-sheet structures increasingly used to move construction costs off corporate balance sheets.

Against that backdrop, a dozen loans being re-marketed simultaneously reads differently than it would have a year ago. It is no longer routine portfolio housekeeping; it is a data point in a live debate about whether the sector's financing has outrun its economics.

Selling Into a Selloff

The timing is unlikely to be a coincidence. The loan sale comes after a brutal stretch for AI-linked equities: this week's Democratic sweep in the US midterms fanned fears of tougher regulation of the technology, triggering a selloff that hit chipmakers and so-called neocloud companies particularly hard.

The repricing evaporated roughly $3 trillion in stock market value from S&P 500 companies, Reuters reported, citing the broader market slide. Data center operators took the worst of it: Applied Digital, Cipher and IREN each tumbled more than 14% in the rout.

Against that backdrop, marketing $500 million of data center debt all at once — rather than gradually — reads as a bank choosing speed over price. Auctions of this kind are typically used when a seller wants a clean, competitive outcome on a defined timetable, and Reuters noted that the bank is taking bids for all twelve loans simultaneously.

Why It Matters

The AI boom has been financed on an enormous scale, and questions about who ultimately bears the risk have grown louder as the industry's borrowing has outpaced its revenue. Banks, private credit funds and bond investors have all piled into data center lending, drawn by long-term leases from investment-grade tenants.

But the midterm elections changed the political calculation. Voters rejected AI data center projects in nearly every ballot measure that touched the issue, and the incoming Congress is expected to scrutinize everything from utility rate-shifting to the debt structures behind the buildout. A European bank reducing its exposure weeks after those results is the kind of move that risk desks notice.

It is also a test of market depth. If the November 19 auction draws strong bids near par, it will suggest appetite for data center debt remains intact despite the equity rout. If bids come in light or at steep discounts, ING's quiet sale could be remembered as the moment AI financing anxiety showed up in credit markets, not just on trading screens.

The presence of a Stargate-linked loan raises the stakes further. Stargate's backers — OpenAI, SoftBank Group and Oracle — have framed the venture as essential national infrastructure, and any indication that lenders are marking down exposure to it, even at the margin, will feed the debate over whether the AI buildout's financing is as durable as its ambitions.

Either way, the sale offers a rare public price discovery moment for a market that has mostly grown in private — and the result will be watched far beyond the construction sites it finances and into every boardroom backing the AI buildout.

Stay Ahead of AI

For continuous coverage of AI business, finance and infrastructure, follow AI Buzz Wire.

Read more AI news at aibuzzwire.news.