数据中心REIT回归基本面：非REIT投资者如何理解回报

Data Centers (DLR, EQIX): REIT-urning to basics - contextualizingreturns for the non-REIT investor Investor interest in data center REITs has expanded rapidly in the AI boom, attracting anincreasingly broad pool of capital beyond the traditional REIT crew. Non-REIT investorspuzzle over the right way to think about these companies—both how to contextualize themagainst one another and against alternate investment opportunities. Madison Rezaei+1 917 344 8622madison.rezaei@bernsteinsg.com Nancy Wu+1 917 344 8545nancy.wu@bernsteinsg.com Data center REITs often screen poorly on traditional metrics such as return on assets orearnings-based multiples, largely due to their capital intensity, long development timelines,and accounting treatments.Importantly, these apparent distortions do not reflectweak business fundamentals, but rather the limitations of applying short-durationfinancial metrics to assets that are designed to operate and generate cash flowsover multi-decade horizons. Like most forms of real estate, data centers are characterized by exceptionally long usefullives (in some cases, 50+ years!) and significant upfront investment. These assets requirelarge-scale development expenditures and years of construction before generatingrevenue, but once stabilized, they can deliver durable and predictable cash flows fordecades. As a result, traditional measures can understate their true economic value. Withinthe REIT framework, Adjusted Funds From Operations (AFFO) provides the most relevantlens, as it best approximates distributable cash flow. Yet even AFFO does not fully capturethe development-driven nature of the business, particularly for companies that continuouslyreinvest capital into new capacity. As an additional metric, today we are looking atyield on cost. This approach evaluatesstabilized returns relative to the original development investment, offering a clearer viewof asset-level economics. Leading data center REITs generate attractive yields on cost—approximately 11% for Digital Realty and 26% for Equinix (wholesale v. retail)—well above those typically achieved by private market developers, which tend to clustercloser to the HSD range. These elevated returns help explain how REITs can sustain growthwhile still supporting shareholder distributions, even when headline valuation multiplesappear demanding. Additionally, we flag the importance of capital allocation strategies in sustaining thesereturn profiles. Digital Realty has been particularly sophisticated, leveraging JV structures,a closed-end fund, and now an open-end strategy (via its recent acquisition of ColumbiaCapital) to monetize stabilized assets and reinvest in higher-yield projects. Equinix takesa different approach: its higher-margin interconnection and retail colocation model drivesa repeatable value-creation cycle. It has also raised a $15B JV-backed developmentfund (with GIC and CPP) to support over 1.5GW of xScale capacity, extending its balancesheet while preserving capital for higher-return retail deployments.In this context,elevated public market multiples appear more justified, as investors are effectivelyunderwriting a long-duration growth model anchored by attractive developmenteconomics and reinvestment opportunities.We maintain our Outperform ratings onboth EQIX ($1,222) and DLR ($232). BERNSTEIN TICKER TABLE INVESTMENT IMPLICATIONS We value DLR on a Price to Adjusted Funds From Operations (AFFO) per share multiple. Our $232 price target is based on 27xour 2027E AFFO per share of $8.52. We value EQIX on a Price to Adjusted Funds From Operations (AFFO) per share multiple. Our $1,222 price target is based on25x our 2027E AFFO per share of $48.63. DETAILS WHAT DO STABILIZED RETURNS LOOK LIKE? We evaluate stabilized asset returns for Digital Realty (DLR) and Equinix (EQIX) using two approaches:ROA—defined as totalstabilized NOI divided by gross PP&E—andyield on cost—defined as estimated stabilized net income divided by average netPP&E. While differences in asset definitions, capitalization policies, and depreciation conventions can introduce some variabilityin the denominator, these measures together provide a consistent directional view of underlying asset productivity. Across bothframeworks, we observe a persistent return premium at EQIX. We believe differences inpricing model and customer mixrepresent a primary driver of this divergence. Equinix is moreheavily weighted toward retail colocation, which is typically characterized by smaller deployments, higher unit pricing, andstructurally higher margins. In contrast, Digital Realty has greater exposure to wholesale colocation, including powered shellofferings, which tend to involve larger-scale deployments at lower unit margins but benefit from longer contract durations andgreater revenue stability. This fundamental difference in go-to-market strategy likely contributes meaningfully to the higherstabilized asset returns observed at EQIX. We also observe a materia