Introduction
The value of a long data center lease rests almost entirely on one assumption: that the tenant will pay rent for the next ten to fifteen years. That makes tenant credit the single most important variable in the sector, and it produces a striking paradox. The counterparties driving the demand surge have among the strongest credit profiles of any tenants in real estate, yet the sector is more exposed to a handful of names than almost any other property type. Both halves of that statement are true at once, and a banker has to hold them together.
The Counterparty Universe
The hyperscalers are a small club. The names that dominate data center leasing are Microsoft, Amazon, Alphabet, Meta, and Apple, with Oracle increasingly in the mix. Their credit ratings sit at the very top of the corporate spectrum, several of them stronger than most sovereign governments.
| Tenant | S&P | Moody's |
|---|---|---|
| Microsoft | AAA | Aaa |
| Apple | AA+ | Aaa |
| Alphabet | AA+ | Aa2 |
| Amazon | AA | A1 |
| Meta | AA- | Aa3 |
- Investment-Grade Tenant
A tenant whose credit is rated BBB-/Baa3 or higher by the major agencies, signaling a low probability of default. The hyperscalers sit far above that threshold, near the top of the scale, which is what lets data center landlords finance long leases on terms closer to corporate bonds than to traditional property loans.
That credit quality is backed by extraordinary financial scale. Combined hyperscaler capital spending is forecast to exceed $600 billion in 2026, a roughly 36% jump over 2025, with Amazon, Microsoft, Alphabet, and Meta each spending more than $100 billion individually, and about three-quarters of that tied directly to AI infrastructure. These are tenants that can write multi-billion-dollar lease commitments and fund them out of operating cash flow, which is precisely why their leases support the financing that the data center demand surge runs on.
The Other Side: Concentration Risk
The strength of these tenants is also the sector's defining vulnerability. Because the majority of new capacity is pre-leased to a few hyperscalers, a data center portfolio's revenue can depend on a small number of counterparties to a degree that would be unacceptable in office or retail. Moody's has explicitly flagged that the pre-leasing wave, while it reduces the risk of empty buildings, is steadily increasing counterparty concentration risk.
That dynamic is what separates careful underwriting from headline chasing. A portfolio heavily pre-leased to hyperscalers looks pristine on a credit screen, but the analysis has to go further: How many distinct tenants are there? How staggered are the lease expirations? Could the space be re-let, and to whom, if a tenant walked. These questions matter most for the largest single-tenant deals at the hyperscale end of the market, where one lease can represent an entire building, and far less for diversified retail colocation. They also flow straight into how the underlying lease structure allocates renewal and termination risk.
Hyperscaler credit is a genuine strength, near-sovereign quality backing a decade or more of contracted rent, but the same pre-leasing that delivers it concentrates a portfolio's fate in a few hands. A strong answer names the universe, acknowledges the exceptional credit, and then makes the more sophisticated point: the real risk in data centers is concentration and re-leasing, not default.
Free Interview PDF Guide