DIGITAL REALTY TRUST, INC. (DLR) stock intrinsic value fair value estimate & owner earnings analysis

Digital Realty owns and operates data centers: buildings with the power, cooling and network connections that cloud providers, software companies, social media platforms, banks and telecom carriers rent to house their servers. It has roughly 2.9 gigawatts of IT capacity in more than 55 metro areas on six continents, about 770 megawatts under construction, and land and power for more than 5 gigawatts more. About 52% of 2025 revenue came from the United States (Northern Virginia, Dallas, Chicago, Portland and others) and 48% from abroad, mainly Europe, with Johannesburg and Singapore also material. Revenue was $6.1 billion in 2025. Customers are concentrated: the largest (a Fortune 50 software company) is 11.7% of recurring revenue and Oracle is 9.0%; the top 20 together are 51%, with a weighted remaining lease term of about six years. How we measured earnings. We start with operating cash flow, strip out working-capital swings, subtract stock compensation and preferred dividends, and subtract the capital the company must spend just to keep its existing buildings running. Digital Realty spent $3.2 billion on its properties in 2025, but its own breakdown shows that about $2.5 billion of that went to building new capacity; only about $372 million was recurring maintenance and small improvements. We treat only that $372 million as the cost of staying in business. The new construction is growth investment, paid for largely by selling new shares, borrowing and selling stakes in finished buildings to joint-venture partners. We don't ignore that funding: every share issued to pay for growth lowers earnings per share, and we measure everything per share. On this basis owner earnings were about $5.37 per share in both 2025 and 2024, and between $4.92 and $5.99 in every year since 2016. That flat record matters most. Revenue almost tripled from 2016 to 2025 and total owner earnings rose about 2.5 times, but the share count rose from about 151 million to 348 million, so each share's earnings barely moved (about 0.8% a year). So far the development machine has grown the company much faster than it has grown value per share. We set our starting point at $5.19 per share, the median of 2020 to 2025. That period starts with the Interxion merger, which created today's global business, and also includes the 2022 Teraco purchase in South Africa. Base case, about $91 per share. Owner earnings per share grew only 0.1% in 2025, but we let them grow 3% a year for 2026 to 2028, then 1.5% a year through 2035. Two things in the filings support the faster start. First, about $4.7 billion of data-center construction is underway, and the company says it expects to deliver that space within 12 months. Second, in the first quarter of 2026 revenue rose 16% and operating cash flow rose 33% while the share count rose only 2.5%. In 2025 the buildings owned for the full period grew revenue 6.1%, helped partly by a stronger euro. Against that, the company's debt is cheap and has to be refinanced: $18.6 billion at an average rate of just 2.90%, with about $10.4 billion due by 2030. We assume that debt is rolled over about 1.5 percentage points higher, which takes away roughly $0.80 per share a year by 2035. After 2035 we hold earnings flat rather than growing them. The buildings, their power connections and the dense customer and network communities inside them have kept earning for more than twenty years, and maintenance spending is already deducted. We assume no growth beyond that. Bear case, about $47 per share. Per-share earnings stay flat before interest costs, as they did over the past decade. Debt is refinanced 2.5 percentage points higher. After 2035 earnings fade to zero over 20 years. This reflects a real risk: recurring maintenance spending of about $370 million a year is small next to the physical depreciation of the buildings and equipment. Much of that equipment (cooling, power systems, generators) wears out over 15 to 20 years, and older sites may need expensive refits that the company counts as development rather than maintenance. Customer concentration adds to the downside: two customers make up about 21% of recurring revenue, and several large tenants have less than three years left on their leases. Bull case, about $118 per share. AI and cloud demand keeps new capacity leasing quickly, per-share earnings grow 5% a year through 2030 and 3% a year through 2035, debt reprices only 1 percentage point higher, and built-in rent increases lift earnings about 1% a year for another decade before leveling off. We did not subtract the company's debt from these values, because interest is already paid out of the cash flows we measure. Leverage shows up instead in the refinancing drag and in the wider bear case. We also left out the large gains Digital Realty books when it sells or contributes buildings to its partnership funds, often $600 million to $1 billion a year. That money pays for new construction, so we treat it as part of the growth engine, not as earnings. If the mechanical approach of treating most construction as upkeep were used, owner earnings would be deeply negative. We think that would misread a developer whose older buildings grew profits in 2025 on maintenance spending alone. What would make this estimate wrong: on the upside, if new construction starts adding to per-share earnings, rather than just growing the company, for several years in a row. On the downside, if a large tenant such as Oracle or the Fortune 50 software company cuts back, if refinancing costs rise faster than we assume, or if older buildings need heavy refits that end up counted as growth spending. Because returns depend on capital markets, customer concentration and rates, the margin-of-safety price is around $58, well below our base value.