Artificial intelligence has become the loudest argument for data-center real estate, and the data-center REITs are the public-market way to collect rent on that buildout. Marc Lichtenfeld’s Oxford Income Letter pitch dresses this story up as a “SpaceX royalty” play, but the substance underneath is a more ordinary and more credible one: AI needs physical buildings, and someone has to own them. The details are in our Best Data Center REITS: What You Need to Know article.

Why AI Needs Real Estate

Every AI model, from a chatbot to a video generator, runs on servers inside a data center. Training and running those models consumes enormous amounts of computing, which means more servers, more power, and more square footage. The cloud providers racing to build AI capacity do not always want to own every building themselves, so they lease space from data-center operators instead.

That leasing demand is the tailwind behind data-center REITs. The two blue-chip names in the sector, Equinix and Digital Realty, are the landlords that many of the largest tech companies rent from, and their facilities are the physical layer of the AI buildout. Neither is a pure AI company, but both sit directly in the path of the spending.

The Power Bottleneck

The less glamorous part of the AI data-center story is power. Data centers are electricity-intensive, and the newest AI facilities need far more power per square foot than older ones. Getting that power, from grid connections to backup generation, has become the real constraint on growth in many markets, and it is a bigger limit on supply than the cost of the building itself.

For the REITs, that cuts both ways. Power constraints slow down new supply, which supports rents for the facilities that already exist. But they also mean that growth is gated by things outside a landlord’s control, like grid capacity and permitting. Investors betting on the AI buildout should understand that the bottleneck is power, not just demand.

The Dividend Is the Payoff

Because data-center REITs are structured as REITs, they pass most of their taxable income to shareholders as dividends. That is the income angle of the trade, and it is what the Oxford Income Letter promo repackages as “royalties.” The reality is more modest: these are low-yield, growing-dividend stocks, typically in the 2 to 3 percent range, whose appeal is the combination of a growing payout and a long runway of demand.

Our data center REITs explainer breaks down that structure and the 90 percent distribution rule behind it, and our Equinix stock piece profiles the largest name in the sector.

The Honest Frame

The AI data-center thesis is legitimate, and it is a better reason to own these stocks than the SpaceX royalty gloss the promo wraps around them. Demand is real, supply is constrained by power, and the landlords collect rising rent either way. The full breakdown of how the promo reframes this as a SpaceX play is in our SpaceX royalty portfolio teardown.

The Risks Nobody Talks About

The AI data-center story has real risks, and most of them are about power and timing. The first is electricity. Data centers are power-hungry, and the newest AI facilities need far more power per square foot than older ones, which has turned grid access into the binding constraint in many markets. A landlord can buy land and pour concrete, but if the local grid cannot deliver the megawatts, the facility cannot be filled. That constraint supports rents in the short run but limits how fast the sector can grow.

The second risk is oversupply. The same AI demand that justifies today’s buildout is attracting enormous capital, and the sector is building at a pace that assumes the demand keeps accelerating. If the AI buildout pauses or slows, a wave of new supply could meet weaker demand, pressuring rents and occupancy. Data centers have been resilient through past cycles, but they are not immune to a supply-demand mismatch.

The third risk is concentration. The sector’s biggest tenants are a handful of cloud providers, and a pause in their spending shows up quickly in leasing. The dividend side carries its own nuance: these REITs yield only a few percent, so the income they produce is modest unless the position is large. Investors who come for the royalty are really getting a low-yield growth dividend, and the growth, not the current payout, is what justifies the premium valuation the sector commands. That is the honest trade here: you are buying the physical layer of the AI buildout, with all the power constraints and supply risk that entails, in exchange for a modest but growing dividend and a long runway of demand. Watching those two dials, power and supply, is the best way to separate a durable thesis from a crowded trade.

Ready to see the research? Click here to access Marc Lichtenfeld’s report.

NewsletterVetter is an independent publication. We receive compensation from some of the services we review through affiliate links. Nothing on this site is investment advice. Always do your own research.