The market for private bonds financing AI data centers, raised through off-balance-sheet joint ventures by the five major hyperscalers, has reached approximately $90 billion. However, the fixed-income market is making a perilous pricing error: investors are substituting the brand credit of the guarantor for a careful evaluation of contract terms. Fitch Ratings is now exposing this illusion through differentiated credit assessments.

As Fitch senior director Anubhav Arora told Bloomberg, investors are focusing solely on whether the parent company providing the lease guarantee is Alphabet (GOOGL) or Meta Platforms Inc, overlooking the fact that "rent doesn't start until the project is complete." If data center construction is delayed, bondholders face a cash flow gap. Notably, Oracle (ORCL), the financially weakest among the five giants, secured Fitch's highest credit score in this sector to date for a transaction where it committed to paying rent from a specified date, regardless of project completion.

Goldman Sachs credit analysts note investors' tendency to treat JV project bonds as a homogeneous asset class. However, Citigroup's analysis of three JV projects for cloud operator FluidStack, all backed by Alphabet, demonstrates that credit outcomes can range from investment-grade down to near-junk status for the same guarantor. This divergence hinges on specific guarantee terms, such as rent discounts, tenant exit rights, and capped-price construction contracts.

Three Projects, Three Different Outcomes

The three FluidStack data center JV projects issued nearly $11 billion in bonds, all guaranteed by Alphabet via lease agreements, yet received three distinct ratings. According to Citigroup analysts, the first project features a full lease guarantee from Alphabet, meaning Alphabet would assume the lease directly if the cloud operator defaults. The second project's guarantee includes a 25% rent discount. The third project's rent formula is linked to operational costs. While the latter two guarantees can cover debt service, they offer significantly weaker protection than the first.

More critical differences lie in the construction contracts and exit clauses. Only one of the three projects has a capped-price construction contract. Another includes a clause allowing FluidStack and Alphabet to exit the lease if construction delays exceed six months. For bondholders, this means that if a project stalls, they could lose rental income and potentially have no party to pursue for recourse.

Ultimately, only the HUT 8 DC LLC project in St. Francisville, Louisiana, valued at $3.25 billion, secured an investment-grade rating. This project uniquely combines a full Alphabet lease guarantee, a capped-price construction contract, and no tenant exit rights.

Oracle's Counterintuitive Top Score

Among the five major hyperscalers, all except Oracle—Alphabet, Amazon.com Inc, Meta Platforms Inc, and Microsoft Corp—enjoy robust cash flows. Yet, it is Oracle, the weakest financially with a rapidly widening CDS spread, that secured Fitch's highest credit score in this field for its Michigan data center project, thanks to an unconditional rent payment guarantee.

Arora's analysis highlights a core contradiction: investors look at the guarantor's brand, while rating agencies scrutinize the details of the terms. When a project combines "tech giant backing" with "tenant exit rights," the former can create a potentially fatal illusion of safety for investors.

The Countdown to Homogenized Pricing

Despite significant variations in terms between JV project bonds, financing costs for most projects remain highly similar, and bond price movements are closely correlated. Goldman Sachs credit analysts anticipate greater yield divergence as the importance of protective clauses—or the lack thereof—becomes increasingly apparent.

Supply-side pressures are amplifying this divergence. The five giants have issued nearly $200 billion in corporate bonds this year, and six AI hyperscalers have collectively raised about $244 billion, more than double last year's amount. Goldman Sachs chief credit strategist Amanda Lynam points out that the combined additional issuance capacity for these five in the USD investment-grade bond market is only about $510 billion—a minuscule fraction relative to Goldman's estimated total capital expenditure need of $5.8 trillion.

This indicates that JV bonds will remain an indispensable financing tool. Tech giants are already tapping euro, Swiss franc, and pound sterling bond markets and utilizing private credit channels through firms like Apollo Global Management. Concurrently, data center demand is so robust that developers are facing real-world constraints like labor shortages. Arora notes this could give landlords greater bargaining power and means deal terms will remain highly bespoke rather than converging toward standardization.

For fixed-income investors, the core question is no longer "is AI infrastructure worth investing in?" but rather "on which side of the terms are you positioned?"—The identity of the guarantor and the conditions of that guarantee are evolving into two distinctly different risk exposures.