Stack Infrastructure, the data-center arm of the private-credit giant Blue Owl, is seeking a loan of about $5.9 billion, according to a recent report. The deal, if it closes, would rank among the larger single financings in a market that has become the engine room of the artificial intelligence build-out.
Blue Owl has made itself central to that market. Through Stack and a string of joint ventures, it has become one of the biggest private financiers of the data centers AI needs, part of a wave of borrowing that now runs into the hundreds of billions. The new loan follows quickly on the last. Stack raised $2.1 billion as recently as February, and the jump to nearly $6 billion in a matter of months is a measure of how fast the appetite for capacity, and for debt, is growing.
The structure is characteristic of the moment. Rather than fund data centers from cash, the companies building them increasingly borrow against the assets and the long-term contracts that anchor them, a model that spreads the enormous cost across lenders hungry for yield. Blue Owl is no bystander in that shift. It has helped arrange tens of billions in financing for hyperscale projects, including large facilities tied to Meta, putting private credit where traditional banks have grown more cautious.
The scale is what has changed. Individual AI campuses now cost tens of billions to build, sums that outstrip what any single bank wants to lend, which has handed private-credit firms like Blue Owl a central role they did not hold a few years ago. The caution has a reason. Data-center debt is only as safe as the demand that underpins it, and a few high-profile wobbles, including a soured Oracle-linked deal, have reminded lenders that the AI boom carries real credit risk beneath the enthusiasm.
Oracle needed PIMCO to anchor a $16 billion financing after banks pulled back, and infrastructure funds from Blackstone to Brookfield are writing cheques that would have looked outlandish two years ago. The lenders are betting on a simple proposition. Compute is scarce, the companies renting it are among the richest on earth, and the contracts behind these buildings can run for a decade, which makes the loans look, on paper, like some of the safest bets in infrastructure.
The risk is that the proposition is circular. Much of the demand comes from AI firms still burning cash, and some of the debt is being backstopped by chipmakers with their own reasons to keep the build-out going, which knits the industry's fortunes tightly together. Blue Owl's size gives it room the banks lack. As a private-credit manager, it can hold loans that regulated lenders would balk at, and it has leaned into that advantage as the AI-infrastructure market has swollen.
The $5.9 billion figure, for now, is based on reporting rather than a closed deal. Terms, pricing, and the exact assets behind it will decide how the market receives it, and large data-center loans have both sailed through and stumbled in recent months. What is not in doubt is the direction. The money required to build AI has outgrown balance sheets and is being met, more and more, with borrowed capital, and Stack's latest raise is another rung on a ladder the whole industry is climbing.
Each of these loans is a wager on the same future. If AI demand holds, the debt is cheap money against a durable asset; if it cools, the buildings and the borrowings will both look heavier than they do today. For Blue Owl, the calculation is one it keeps making. The firm has decided that financing the picks and shovels of the AI rush is the safer way to profit from it, and a $5.9 billion loan is its latest move to stay at the front of that queue.
To understand the significance of this loan, one must look at the broader landscape of AI data-center financing. Over the past two years, the demand for compute power to train and run large language models has exploded, driven by companies like OpenAI, Google, Microsoft, and Anthropic. These firms require massive clusters of graphics processing units (GPUs) and specialized hardware, typically housed in purpose-built data centers. The capital expenditure required is staggering: a single hyperscale data center can cost between $1 billion and $10 billion, with some AI-focused campuses exceeding $20 billion.
Traditional bank lending has struggled to keep pace. Regulatory constraints, capital adequacy requirements, and risk aversion have made many banks reluctant to underwrite billion-dollar loans for speculative assets. This gap has been filled by private credit firms, which can move quickly and structure bespoke debt facilities. Blue Owl, with over $150 billion in assets under management, has been at the forefront. Its data-center arm, Stack Infrastructure, was formed through a series of acquisitions and partnerships, including a joint venture with DigitalBridge and other investors.
Stack's business model is to develop, own, and operate data centers that are pre-leased to hyperscale tenants. The long-term leases—often 10 to 15 years—provide predictable cash flows that support high levels of debt. The $5.9 billion loan being sought is likely secured against a portfolio of such data centers, possibly including facilities that will house infrastructure for major cloud providers or AI startups. The loan is expected to be structured as a first-lien senior secured facility, with interest rates tied to the Secured Overnight Financing Rate (SOFR) plus a spread that reflects the perceived risk.
Market participants are watching closely. If successful, this would be one of the largest private credit deals ever in the data-center sector. It would also signal that lenders are comfortable with the leverage levels and tenant creditworthiness. However, there are cautionary notes. Last year, a large data-center loan backed by Oracle faced difficulties after the company's debt rating was downgraded, forcing a restructuring that saw PIMCO step in. Similarly, some smaller data-center operators have defaulted when lease commitments fell short of projections.
The growth of AI itself introduces a unique risk. Many of the companies that are the primary consumers of data-center capacity are not yet profitable. OpenAI, for example, is expected to spend over $5 billion on compute alone in 2024, while generating less than half that in revenue. If the pace of AI adoption slows or if fundamental breakthroughs plateau, the demand for new data centers could decline rapidly, leaving lenders with underutilized assets. Furthermore, chipmakers like Nvidia are themselves facilitating data-center construction by offering financing or equity stakes, which ties their fortunes even more tightly to the boom.
Despite these risks, the momentum shows no signs of abating. In the first quarter of 2025, global data-center investment reached a record $35 billion, with private credit accounting for roughly 40% of that financing. Blue Owl's $5.9 billion loan would be one of the quarter's largest transactions. The firm has also been active in other corners of the AI infrastructure market, including fiber networks and renewable energy projects to power data centers.
Beyond the immediate deal, the broader trend raises questions about the stability of the financial system. While private credit is less regulated than banking, it also is less transparent. Many of these loans are held in funds that are not marked to market, so the true value can be obscured. Regulators have begun to take notice, with the Federal Reserve and the Bank of England both issuing warnings about the concentration of risk in private credit. However, no immediate action is expected, as policymakers are wary of stifling innovation in AI.
For Stack and Blue Owl, the $5.9 billion loan is a bet that the AI data-center boom will continue for at least the next decade. The company's strategy relies on being able to refinance debt when needed, and the current environment of low defaults and high demand for yield is favorable. But as with any debt-fueled boom, the cycle can turn. The key metric to watch will be the occupancy rates of new data centers and the financial health of their tenants. If those remain robust, then Stack's loan will be seen as a prudent capital move. If not, it could become a cautionary tale.
Source:TNW | Investors-Funding News