Investing in Data Centers

Posted

It seems like such a simple question. Who owns data centers?

Ownership structures in the digital economy are more varied than might appear on the surface. While the largest computing and cloud service providers, such as Apple, Amazon (AWS), Microsoft (Azure) and Google (Cloud)—also known as “hyperscalers”—do own and operate a significant portion of their global infrastructure, they are increasingly partnering with third-party developers and investors, including real estate investment trusts (REITs), to expand capacity and deploy capital quickly and efficiently. This article provides a guide to how the most prominent strategic and financial players are engaging in this sector.

The Variety of Investment Approaches

Hyperscalers
Hyperscalers at the largest scale often own core campuses in strategic global regions and carry most if not all of the capital cost on their own books. Publicly reported examples include Microsoft’s data center campus near San Antonio, Texas, Google’s facilities in The Dalles, Oregon, and Northern Virginia, and Meta’s Kuna data centers in Kuna, Idaho. Most hyperscalers own and operate a dozen or more of their own data center facilities in locations around North America. These companies acquire land (often without leveraging debt), self-fund construction costs, hire employees, and manage and operate these data centers to support their own cloud computing business lines; the economics of hyperscale cloud computing services make these major capital investments worthwhile for these large companies.

This model is less common among smaller (non-hyperscaler) operators. The capital requirements involved in acquiring land and in building and energizing the facility are significant, and non-hyperscaler operators may not have the financial capacity or risk tolerance to pursue direct ownership.

REITs, Developers and Infrastructure Funds
Many large-scale data centers and campuses are owned by infrastructure funds, REITs or private equity-backed developers who specialize in developing, owning and operating data center assets. These data center landlords lease space to hyperscalers, enterprises and other cloud service providers. Though hyperscalers do own campuses and facilities of their own (as described above), many still lease data center capacity from data center-specific REITs (like Equinix or Digital Realty), or partner with developers to build on a sale-leaseback model, which provides the developer/landlord with guaranteed rental income to help mitigate the large financial risk of data center development. REITs and developers on the other hand specialize in acquiring and constructing these facilities, and operators are often happy simply to lease the completed facility.

These arrangements often include long-term, triple-net leases that provide steady, predictable cash flow for the facility owners and reduced capital costs for operators. A hyperscaler or other operator leasing a data center will typically pay a base annual rent, and will reimburse the landlord for all (or a pro rata share, if there are multiple tenants or buildings within a campus) of the landlord’s operating costs, such as maintenance, security, cleaning, landscaping and other management activities. Leasing rather than owning gives an operator flexibility to more rapidly adapt to market and technological changes, tying up capital for only a defined lease term rather than an extended period of ownership, and in some cases providing rights to terminate a lease early. Leasing also helps operators scale up more quickly—when additional capacity is needed, leased space can be brought online much more quickly than newly constructed space.

Other Investors
It is possible for others to invest in data centers, but options depend on financial capacity, risk appetite and access to institutional networks. Some avenues through which interested investors (including individuals) might gain exposure to the data center market are:

  • Public Markets (REITs). This is the most accessible route. Publicly traded data center REITs like Equinix (EQIX), Digital Realty (DLR) and newer entrants like CoreSite (now part of American Tower) offer direct exposure to the sector. These REITs provide liquidity, dividends and built-in diversification across geographies and tenants.
  • ETFs and Mutual Funds. While there aren’t many mutual funds focused exclusively on data centers, several tech and infrastructure ETFs include heavy allocations to REITs and hyperscaler-linked real estate. Examples include the Pacer Benchmark Data & Infrastructure Real Estate ETF (SRVR).
  • Private Equity and Infrastructure Funds. Access to private equity or other infrastructure funds that invest in data centers is more limited. These types of funds often require minimum investments in the seven-figure range and are typically open only to institutional investors and high-net-worth individuals. However, some private wealth platforms are beginning to offer limited access to infrastructure-focused funds.
  • Direct Investment or Syndication. A more niche option is to invest directly in regional operators or through syndication platforms offering co-investment opportunities in smaller to medium-sized data center facilities. These are higher risk but potentially higher return, often suited for investors with both domain knowledge and capital.
  • Investment in Power Suppliers. One interesting way to participate in the data center phenomenon is to invest in utilities and independent power producers that have a particular specialty in projects to supply energy for power and cooling for data centers.

The Data Center Capital Stack: Equity, Debt and Hybrids
Data centers are capital-intensive to build and operate. A hyperscale facility can cost hundreds of millions of dollars, and even smaller regional data centers often require tens of millions in up-front capital. Data center developers may seek financing for these projects from a range of sources, and may mix and match these sources, depending on the type and scale of the facility as well as regional leasing activity and demand.

Private Equity (PE) and Infrastructure Fund Equity
PE and infrastructure investors have poured capital into both developers and operators, funding greenfield builds and platform acquisitions. For example, Microsoft and BlackRock are publicly reported to have teamed up with technology-focused private equity firms to create a Global AI Infrastructure fund which aims to raise $30 billion for the construction and energizing of data centers; xAI and Nvidia have since joined the fund. Investment funds of this kind can deploy capital relatively quickly when opportunities arise, and can leverage debt resources on a project-by-project basis. Once a project is completed and operational, the fund (or other investment vehicle) may seek to sell the project at a profit.

REIT Equity and Debt
Data center REITs access both public equity and debt markets to finance new developments or acquisitions. Their access to capital is one reason hyperscalers increasingly partner with REITs. REITs acquire or invest in the land and facility directly, either independently or by entering a joint venture with an operator. REITs may work with developers to construct new projects or may acquire already-completed facilities and lease them out to operators. As with PE and infrastructure investors, REITs can raise funds from a variety of sources and deploy those funds quickly when opportunities arise. Unlike hyperscaler-owned projects, REITs typically have a specific investment period or horizon, after which they will look for exit opportunities rather than continuing to own a project or facility indefinitely.

Corporate and Project-Level Debt
Securing long-term debt to finance large infrastructure projects such as data centers is common, particularly in stabilized assets with secure tenant leases. For example, a developer may negotiate with a hyperscaler to enter an agreement to lease a yet-to-be-completed data center and then provide a lender with the finalized lease to make the lender comfortable that the project will provide sufficient cash flow as a means to secure more favorable loan terms. Because data centers tend to have long occupancy periods and low turnover as compared to other asset classes, cash flows can be easier to predict, and vacancy risks can be minimized. This in turn attracts potential lenders to underwrite these sizeable projects. Alternatively, hyperscalers themselves can use debt to defray their own capital outlays for new projects; the lender can rely on cash flow projections of the hyperscaler itself rather than potential lease income to gain comfort lending money in these cases.

Debt terms can range from traditional bank loans to infrastructure debt and even green bonds when sustainability targets are involved.

Hybrid Approaches
The investment options are also ripe for consideration of hybrid approaches. While some states are straining under the power and other requirements of data centers, other localities are looking to lure them in with tax abatements, grants and other incentives. Government users will increasingly demand data center capacity of their own in the 21st century. These alternative approaches to investment may entail public-private partnerships and joint ventures and other forms of collaboration and equity, convertible debt and conventional debt instruments.

The sheer size, diversity and sophistication of the modern data center can test the most experienced of investors and most robust of capital stacks. For those contemplating investment in data centers and the related digital economy, knowing where to start, what to do on the way, and what comes next after you get there requires its own substantial stack of sector-specific legal, business and regulatory guidance.


RELATED ARTICLES

Powering Data Centers with Nuclear Generation

Data Centers: A Field Guide

Anatomy of a Data Center