NextFin News - Stack Infrastructure is seeking about $5.9 billion of debt for a new financing package, a sign that the artificial-intelligence buildout is being pushed deeper into the credit market. The amount is large enough to matter on its own, but the more important point is that the market is now treating AI data centers as repeatable, financeable industrial assets rather than one-off mega-projects.
The new loan would extend a pattern already visible in STACK’s own disclosures. The company said in a June 4, 2024 release that it had secured more than $12 billion in financing for its global portfolio. On March 19, 2025, it said it had secured $4 billion in green financing for a 1+GW campus in Stafford, Virginia, as well as campuses in Portland, Oregon, and Toronto, Canada. It also said on Aug. 5, 2024 that it had secured an additional $3 billion of green financing and had then raised more than $15 billion to support its global portfolio. Those figures show a developer that has already built a large, repeatable funding machine around data-center campuses.
Blue Owl sits at the center of that machine. The firm’s digital-infrastructure platform says it is focused on developing, acquiring and owning data centers and other connectivity-related real assets to meet AI and cloud-driven demand. In a June 2026 investor presentation, Blue Owl said its digital infrastructure strategy is part of a broader real-assets platform and highlighted its role as a capital provider for hyperscaler-related projects. That positioning helps explain why a $5.9 billion loan is more than a single transaction: it is part of a broader shift in how the AI economy is being funded.
The immediate market question is not whether AI spending exists. It is who will pay for the land, shells, power, cooling systems and long-dated construction needed to turn compute demand into revenue. Debt is filling that gap because it can bridge the period between capital outlay and contracted cash flow. The first-order effect is obvious: more debt means more projects can get built. The second-order effect is less obvious: once the financing model is standardized, lenders begin to underwrite the AI boom itself, not just the individual borrower.
That is why the headline loan matters beyond the number. The market is not just financing data centers. It is increasingly financing the tempo of AI capacity additions. If that tempo remains strong, the credit market can keep absorbing these deals. If it slows, the same structure can unwind quickly because debt is most forgiving when utilization, tenant demand and refinancing conditions all cooperate.
What The New Loan Says About The AI Credit Cycle
The $5.9 billion figure matters because it is big enough to support a multi-campus or platform-scale capital stack, not just a single building. In data-center finance, that distinction matters. Large loans often sit inside broader development plans that combine land, power commitments, construction phases and long-term tenant relationships. When a lender writes a facility of that size, it is not only financing concrete and steel. It is taking a view on how quickly hyperscale demand can be converted into stabilized cash flow.
STACK’s own history shows how that mechanism works. The company said in June 2024 that it had secured more than $12 billion in financing, then added another $3 billion in August 2024, and then announced $4 billion more in March 2025. Those disclosures are not just fundraising milestones. They show that the company has repeatedly accessed capital at scale, which is exactly what makes the next loan easier to imagine and, in some cases, easier to place. The financing market learns by repetition. Each completed deal reduces the perceived novelty of the next one.
“Blue Owl Digital Infrastructure Fund III seeks to support the global capital demands of hyperscalers, fueled by ongoing investments in data centers and artificial intelligence.” — Blue Owl Capital, fund announcement
That statement captures the core of the trade. The capital is not flowing despite the AI buildout; it is flowing because of it. In the short run, that is a cyclical credit boom. Liquidity is available, investor appetite is high, and lenders are comfortable underwriting large projects against a growth narrative that has not yet broken. But the same feature makes the boom vulnerable. If capital markets tighten, or if one marquee project misses utilization targets, the willingness to fund ever larger debt packages can recede quickly.
The counterpoint is that this is not just a fad. AI workloads require massive amounts of compute, power and cooling, and those needs do not disappear because spreads widen for a quarter. The asset class has become more utility-like, with heavier upfront capex and longer development lead times. That gives the financing model a structural element: a data center is not software, and it cannot be built out of retained earnings alone. Private credit and project finance are becoming part of the operating model for the industry.
That structural reading is strengthened by Blue Owl’s own platform buildout. The firm said in an April 2026 investor presentation that it had $315 billion of AUM, up from $62 billion at public listing, and highlighted $85.1 billion of real-assets AUM and $70.6 billion of credit AUM. Those figures help explain why it has the balance-sheet breadth and product range to keep providing capital for large digital-infrastructure transactions. The market is not relying on a niche lender. It is using a scaled asset manager with multiple credit and real-estate channels.
Still, the strongest counter-thesis is that this is exactly what late-cycle excess looks like. If too much capital is chasing the same AI story, lenders can end up financing capacity that arrives before demand does. A data center can only justify its cost if power, tenants and pricing line up. If they do not, leverage does not create returns; it magnifies the gap between the plan and reality. That is the risk embedded in every oversized loan.
The signal that would disprove the structural-funding thesis is specific: if large AI-linked construction loans begin to come with materially shorter maturities, wider spreads and higher equity cushions across multiple transactions, then the market is no longer treating the category as a durable infrastructure class. That would mean the credit boom was still mostly cyclical, not a lasting change in how the industry is financed.
The Broader Market Is Financing The Boom Itself
The second-order implication is that AI debt is no longer just a byproduct of the equity story. It is now one of the main ways the story gets built in the real economy. Developers need capital before revenue arrives. Hyperscalers need capacity before training and inference demand peaks. Lenders step in because they can be repaid from contracted cash flows if the projects perform as expected. That is the transmission chain: AI demand creates infrastructure demand, infrastructure demand creates financing demand, and financing demand pulls private credit deeper into the cycle.
That chain also explains why the market has begun to tolerate larger and more frequent financings. STACK said it had already raised more than $15 billion by August 2024 and more than $12 billion by June 2024, while Blue Owl’s digital infrastructure platform says it is built around data centers and other connectivity-related real assets. Those facts suggest that the market has already normalized the idea of heavy debt funding for digital infrastructure. The new $5.9 billion package would simply move that normalization further down the curve.
The cyclical-versus-structural answer, then, is split. The near-term credit appetite is cyclical and can reverse. The underlying need for more AI-ready infrastructure is structural. That is why the debt boom can keep expanding even while individual loans become more sensitive to spreads, maturities and tenant quality. The boom is still powered by momentum, but the asset class it is funding is starting to look permanent.
The strongest argument against that view is that the market is extrapolating too much, too fast. If AI capex growth slows, the same lenders that are now eager to fund can become cautious almost overnight. The test is not a philosophical one. It will show up in pricing, maturity, leverage and the willingness of sponsors to add equity rather than more debt. A meaningful widening in those metrics would say the market is moving from enthusiasm to discipline.
For now, the base case is that the credit market keeps funding AI infrastructure because the demand case still looks strong and the projects are still large enough to absorb institutional capital. The upside case is a more standardized market in which contracted revenues and repeatable financing structures reduce uncertainty and lower borrowing costs. The downside case is a demand or funding shock that forces lenders to reprice unfinished capacity and exposes how dependent the boom has become on abundant credit.
The next loans will tell the story faster than the earnings calls will. If the structures stay large, long-dated and tightly underwritten, the market is treating AI infrastructure like a real industrial regime. If they start shrinking or demanding much more equity, the boom is still a trade, not a system.
This is no longer just an AI story. It is a credit story with AI demand attached.
And that makes the financing stack the first place to watch when the boom starts to wobble.
Explore more exclusive insights at nextfin.ai.
