In December 2025, Blue Owl Capital walked away as lead lender on a roughly 1 gigawatt Oracle and OpenAI “Stargate” campus rising in Saline Township, Michigan. Four months later, Bloomberg reported that PIMCO, working with Bank of America, had stepped in to arrange up to $14 billion of debt inside a $16 billion financing package for the same site — with Blackstone reportedly adding roughly $2 billion of equity on top. The building didn’t change. The buyer’s underlying exposure to a single large landlord’s balance sheet did. For anyone signing a multi-year colocation or wholesale lease right now, that swap is the story worth paying attention to, not the gigawatt count.
For most of the colocation industry’s history, the financing behind a facility was a back-office detail that rarely surfaced in a buyer’s decision-making. A provider either had the balance sheet to build or it didn’t, and tenants focused their diligence on power density, Tier ratings, network meet-me rooms, and uptime history. That calculus is changing fast enough in 2026 that ignoring it has become its own risk.
The debt wave behind the AI buildout
Data center capacity is being financed differently than it was even two years ago, and the shift is large enough to show up in global capital markets data. Morgan Stanley projected in 2026 that global AI-linked bond issuance would reach roughly $570 billion for the year, with about $236 billion already priced by the end of May. S&P Global Ratings, in its February 2026 report “Liquidity Outlook 2026: Six Questions, Six Answers,” found that global data center securitization — asset-backed bonds collateralized by leases and facility cash flows — topped $30 billion in 2025, nearly triple the roughly $10 billion issued in 2024.
None of this is inherently alarming. Securitizing a facility’s long-term lease income is a well-established tool in telecom towers, toll roads, and aircraft leasing, and strong investor demand for these deals (several have been oversubscribed) reflects genuine confidence in data center cash flows. But the pace and scale matter to anyone who is a tenant rather than a lender: the capital structure behind your provider is now more likely to include multiple layers of debt, special-purpose financing vehicles, and lenders whose risk appetite can change mid-project.
Why a lender pulling out matters to you as a buyer
The Michigan campus is a useful case study precisely because the financing wobbled in public. Blue Owl’s exit as original lender, followed by months of negotiation before PIMCO and Bank of America assembled a replacement package, is a reminder that “financed and under construction” is not the same as “financing secured.” If you are negotiating a pre-lease or build-to-suit commitment tied to a facility’s delivery date, a lender substitution upstream can translate into schedule risk downstream — even when the eventual deal closes, as this one did.
Separately, Bloomberg reported in late July 2026 that Stack Infrastructure, Blue Owl’s own data center platform, was seeking a $5.9 billion loan, on top of a $2.1 billion raise earlier in the year. That a major sponsor is both pulling back from one project’s debt and simultaneously raising fresh billions for its own platform illustrates how fluid and provider-specific financing conditions have become. It is not a signal to avoid any particular operator; it is a signal that capital-stack questions now belong in buyer due diligence, alongside power availability and SLA terms. The tradeoffs between committing to a single build-to-suit campus versus leasing flexible colocation capacity — covered in more depth in our build-to-suit versus colocation comparison — are increasingly a financing question as much as a technical one.
Need Expert Guidance?
Talk to a Data Center Expert
21+ years of hands-on experience in data center design, operations & infrastructure. Book a quick discovery call to discuss your project.
📞 Book a Discovery CallPrivate credit’s growing share, and the risk it carries
S&P Global’s February 2026 outlook also quantified a broader structural shift: private credit funds extended roughly $146 billion to below-investment-grade borrowers in 2025, compared with about $85 billion through broadly syndicated loan markets — a reversal of the funding mix that prevailed for most of the past decade. Data centers, alongside other AI-infrastructure borrowers, are a significant share of that growth.
The same report flags the obvious follow-on risk: a refinancing wall. U.S. sub-investment-grade debt maturities were projected to climb from $56.6 billion in 2026 to a peak of $215 billion in 2028 — landing squarely inside the window when today’s AI-era data center debt will need to be rolled over or refinanced, against hyperscaler capital expenditure that S&P estimated at roughly $600 billion for 2026, up about 38% year over year. If interest rates, AI demand growth, or credit spreads look materially different in 2028 than they do today, some providers financed aggressively now will face a harder refinancing conversation than their original underwriting assumed. That is a risk that sits with the landlord’s balance sheet first — but it can reach a tenant through renegotiated terms, deferred capacity upgrades, or, in a worst case, a change of ownership mid-lease.
This isn’t a hypothetical confined to Wall Street research notes. A buyer locking in a 10 or 15-year wholesale commitment today is, whether they realize it or not, making a bet that extends past that 2028–2030 maturity wall. If the facility’s debt was priced on today’s favorable spreads and has to be refinanced into a tighter credit market, the provider’s cost base changes — and contracts that lack clear protections around pass-through costs, change-of-control, or service continuity leave the tenant exposed to decisions made entirely outside their control.
Diligence questions worth adding to your next RFP
None of this means buyers should treat heavily-leveraged providers as unsafe, or chase only balance-sheet-funded hyperscalers who can self-finance. It means the financing structure behind a facility deserves the same scrutiny that power redundancy and interconnection already get. A few questions worth adding to vendor evaluation and lease negotiation this cycle: Who actually owns the facility entity you are contracting with, and is it the operating company or a special-purpose vehicle created for this specific project? What happens to your service agreement, SLAs, and expansion rights if that entity is refinanced, sold, or restructured? Is any part of the provider’s debt secured directly against the lease revenue your contract will generate, and if so, does your agreement include protections if the lender forecloses or the asset changes hands? What is the provider’s weighted-average debt maturity profile, and does any material tranche land inside the 2028–2030 refinancing window S&P has flagged industry-wide? None of these questions require a finance background to ask — they just require recognizing that the balance sheet behind your rack space is no longer a background detail.
The nuance: this is still a sign of confidence, not distress
It’s worth being direct about the counterargument, because the hype cycle runs in both directions. Oversubscribed securitizations, tightening spreads on data center ABS relative to 2023, and the sheer willingness of PIMCO, Blackstone, and other blue-chip institutional lenders to step into a deal Blue Owl walked away from all suggest that sophisticated capital still views contracted data center cash flows as a good bet, not a bubble. Debt financing itself isn’t a red flag; infrastructure has been built this way for decades. The actual lesson for buyers is narrower and more useful than “AI data centers are overleveraged”: financing structures are becoming more complex and more consequential to lease stability, so the questions above belong in your process regardless of which provider you’re evaluating.
Capacity planning used to start and end with power, cooling, and location. In 2026, it also has to include a read on who is actually standing behind the facility financially, and what happens to your contract if that changes before your lease term ends. Our Data Center Buyer’s Toolkit walks through a fuller due-diligence checklist, including capital-stack and ownership questions, for anyone evaluating a colocation, wholesale, or build-to-suit commitment in the current market.
Written by
Raajeev Ratra
Data Center Infrastructure Expert | 15+ Years in DC Design, Operations & Project Management
Raajeev is a seasoned data center professional with hands-on experience in hyperscale facilities, colocation design, power & cooling infrastructure, and global DC operations. He shares practical insights to help engineers and IT leaders build better infrastructure.