AI Infrastructure Boom: Challenges and Opportunities for Data Center REITs

The AI revolution has sparked an unprecedented surge in tech valuations, creating hundreds of unicorns and numerous billionaires while adding over a trillion dollars in market value to tech giants like Nvidia, Google, and Meta. This digital gold rush has triggered a massive infrastructure land grab, exemplified by Mark Zuckerberg’s recent announcement of Meta’s plans to build tens of gigawatts of AI infrastructure this decade, eventually expanding to “hundreds of gigawatts or more” at a potential cost of trillions of dollars. Such ambitions highlight the extraordinary scale of investment pouring into AI computing power, as companies race to secure the physical infrastructure needed to support increasingly complex AI systems and their voracious appetite for computational resources.

Despite this boom, traditional data center landlords—the REITs that have historically provided digital infrastructure—are struggling to capitalize on the opportunity. Major players like Equinix, Digital Realty, and Iron Mountain have seen their share prices decline by 13%, 11%, and 16% respectively over the past year, even as the S&P 500 climbed 17%. These companies, which purchase real estate, build data center facilities with supporting infrastructure, and lease to tech companies, should theoretically be thriving in this environment. However, they face structural disadvantages in the current market, where investors believe that chipmakers and tech giants will capture the lion’s share of AI’s economic benefits rather than “mercenary data center developers.” Adding to their challenges, the top 20 data center development contracts for 2025 went to companies other than these established REITs, signaling a shift in market dynamics.

The REIT structure itself poses significant limitations in the AI infrastructure race. Created in the 1960s to democratize real estate investing, REITs avoid corporate income tax but must distribute 90% of taxable income as dividends to shareholders. While this model worked well for data center REITs that emerged around 2000, it’s proving problematic in today’s environment where massive capital investments are required. The dividend obligation leaves REITs with less capital for reinvestment precisely when the industry demands unprecedented levels of upfront commitment. As Andy Cvengros from JLL puts it, as a REIT, “you can’t stick your neck out there” the way competitors are doing, which has “really held them back in a big way.” This structural constraint makes it difficult for REITs to compete in an era of escalating data center bidding wars.

To compete for major contracts, REITs must contend with formidable rivals backed by diverse funding sources. Tech giants like Meta are leveraging existing cash-flow streams to finance their data center ambitions, while private companies such as Vantage and QTS are forming partnerships with investment firms to boost their financial firepower. REITs, however, answer to risk-averse shareholders who resemble pension fund investors rather than tech stock speculators. These shareholders resist investments involving potentially risky debt, which is often necessary for modern data center projects. REITs typically maintain conservative leverage ratios of less than 5x equity, compared to private companies’ 10-15x ratios. This cautious approach makes it challenging to do business with AI players like OpenAI, which has committed to approximately $1.4 trillion in spending without clear funding sources, or CoreWeave with its junk-rated debt. The market’s different reactions to similar announcements illustrate this disparity: When Oracle announced increased data center investments despite an earnings miss, its stock surged 38%; when Equinix made a similar announcement, its share price plummeted 18%.

Beyond financial constraints, data center REITs face additional challenges that complicate their growth prospects. Many large data center contracts include early termination clauses that can be triggered by factors like construction delays—a common occurrence affecting more than half of data center projects in 2025. Power availability represents another major hurdle, as grid capacity limitations create bottlenecks for expansion. As analyst Mark Giarelli noted, “When we have an empty data center shell that isn’t delivering megawatts of power, that will drag returns.” REITs have also been reluctant to venture beyond established markets like Northern Virginia and Los Angeles into the more remote locations where massive AI infrastructure is increasingly being built due to space requirements. This geographical conservatism has contributed to Equinix’s revenue growth falling short of projections and Digital Realty seeing decreased AI-related workload revenue.

Despite these challenges, REITs still have meaningful opportunities in the AI infrastructure landscape. As AI computational demands shift from model building to model running, proximity to users and access to robust fiber networks become increasingly important—advantages that urban data centers possess. Analysts are particularly optimistic about Equinix due to its focus on speed and connectivity, with the company rapidly expanding its xScale business that works with tech giants. The tremendous demand means there’s still room for REITs to claim larger stakes in the boom, as demonstrated by Applied Digital, which transformed from a crypto mining host to an AI data center provider, tripling its stock price and revenue within a year. However, REITs must balance risk management with capital raising to effectively compete. To truly capitalize on the AI infrastructure opportunity, they’ll need to make calculated but decisive bets while satisfying their more conservative shareholders. Otherwise, they risk playing perpetual catch-up in an industry where speed increasingly determines success.