[DLR] Digital Realty Trust Thesis 2026: AI Infrastructure Demand Cycle Drives Hyperscaler Bookings Acceleration
Digital Realty Trust FY2025 revenue ~$5.85B (+5-7%) with FFO/share ~$7.00. AI infrastructure demand cycle drove FY2024 record bookings ~$1B+ (vs $500M historical baseline) — hyperscaler concentration ~80%+ of bookings. Total IT capacity grew from ~2,400 MW FY2021 to ~3,400 MW FY2025 (~42% capacity growth). Development capex elevated to ~$4B+ FY2025 supporting AI demand response (development yield ~8-9% on cost). Hyperscaler customer concentration: Microsoft + Google + Amazon + Meta + Oracle + selected emerging AI customers. Same-store NOI growth ~4-5%. CEO Andrew Power transitioned Jan 2024 (succeeded Bill Stein 2014-2024). Dividend coverage tight at ~93-98% AFFO. FY2026 thesis: AI demand continuation supports record bookings + development pipeline scaling; key risks: AI cycle reset, dividend coverage compression forcing cut, Equinix competitive intensity at hyperscaler accounts.
Key Takeaways
Digital Realty Trust Inc.'s fiscal year 2025 (calendar year ended December 31, 2025) demonstrated the operational benefit of the AI infrastructure demand cycle that has accelerated hyperscaler data center leasing activity to record levels: revenue of approximately $5.7-6.0B (+~5-7% YoY), funds from operations (FFO) per share of approximately $6.80-7.10 on approximately 340M diluted shares supporting Digital Realty's continued shareholder return through the dividend distribution (currently $4.88/share annual yielding approximately 3-4%) plus the substantial capital deployment toward data center development. The strategic identity that distinguishes Digital Realty from peer data center REITs (Equinix as the largest data center REIT covered separately, plus selected smaller emerging data center operators including QTS Realty Trust acquired by Blackstone 2021, Iron Mountain selectively expanding into data center, plus selected smaller) is the deliberate concentration in hyperscaler + selected enterprise data center capacity combined with the global geographic footprint (approximately 310 data centers globally with approximately 3.4GW total IT capacity) that supports the multi-decade hyperscaler customer relationships. The investment thesis for Digital Realty in FY2026 centers on three structural questions: (1) whether the AI infrastructure demand cycle continues supporting record leasing activity through FY2026 — Digital Realty reported record FY2024 bookings of approximately $1B+ in new leasing commitments (versus historical run-rate of approximately $400-500M), with FY2025 trajectory indicating continued elevated bookings supported by hyperscaler AI training cluster + inference deployment requirements; (2) whether the development pipeline (Digital Realty has approximately $3-5B annual capex commitment toward hyperscaler-leased data center development) translates into the targeted operational + financial outcomes as new data centers come online with multi-year hyperscaler lease commitments; and (3) whether the same-store NOI growth (approximately 3-5% reflecting the dual dynamics of multi-year hyperscaler leases at fixed escalators plus selected enterprise customer dynamics) sustains as the portfolio continues maturing.
Digital Realty Trust was founded in 2004 in San Francisco, California as a real estate investment trust focused exclusively on data center real estate ownership + leasing. The 21-year operational expansion built Digital Realty through landmark milestones including the 2004 IPO that established Digital Realty as a publicly traded data center REIT, the multi-decade portfolio expansion through both organic development + selected acquisitions (selected smaller European data center operators acquired through 2010s-2020s expanding Digital Realty's European footprint, the 2017 acquisition of DuPont Fabros Technology for $7.6B that consolidated selected wholesale data center capacity, the 2019 acquisition of Interxion for $8.4B that materially expanded European retail data center capacity, plus selected smaller acquisitions through the recent period), plus the 2024 CEO transition from Bill Stein (CEO 2014-January 2024) to Andrew Power (CEO since January 2024 + previously CFO 2014-2024). The strategic identity that distinguishes contemporary Digital Realty from peer data center REITs is the deliberate dual-format positioning combining wholesale data center capacity (large blocks of contiguous capacity for hyperscaler customers requiring multi-megawatt deployments) with retail data center capacity (smaller cabinet-level capacity for enterprise + selected emerging customers) — the dual format provides revenue diversification + selected emerging cross-selling opportunity though with selected operational complexity versus pure-play wholesale or pure-play retail data center operators.
Business Structure
Digital Realty Trust organizes its business through revenue categories rather than formal segment reporting — the practical organization aligns around customer types + geographic categories.
Wholesale Data Center Capacity (~70% of revenue, ~$4-4.2B FY2025): The hyperscaler customer-focused data center capacity. Sub-categories:
- Hyperscaler customers: Microsoft + Google + Amazon + Meta + Oracle + selected emerging hyperscaler-class customers (selected AI-focused customers including OpenAI through cloud partnerships + Anthropic + selected emerging).
- Lease structure: typical wholesale data center leases are 5-15 year duration with annual escalators of 2-3%, providing multi-year revenue visibility supporting REIT cash flow stability.
- Capacity scale: hyperscaler customers typically require 5-50+ MW of contiguous data center capacity per deployment, supporting the wholesale data center format that Digital Realty operates.
Retail Data Center Capacity (~30% of revenue, ~$1.7-1.8B FY2025): The enterprise + selected emerging customer-focused data center capacity. Sub-categories:
- Enterprise customers: large + medium enterprises requiring smaller-scale data center capacity (selected colocation footprint, hybrid cloud connectivity, selected emerging requirements).
- Selected emerging customers: AI startups, selected fintech operators, plus selected emerging digital infrastructure customers.
- Lease structure: typically 3-7 year duration with selected emerging shorter-term arrangements.
Geographic Distribution (FY2025 estimate):
- North America: ~60% of revenue (US dominant + selected Canadian operations) — Northern Virginia (Ashburn dominant cluster), Dallas, Chicago, Silicon Valley, plus selected emerging US markets.
- Europe: ~30% of revenue — London, Frankfurt, Amsterdam, Paris, Dublin, plus selected emerging European markets.
- Asia-Pacific + Latin America: ~10% of revenue — selected operations across major Asian + Latin American markets.
Key Core Metrics Performance
Revenue, FFO, and Same-Store NOI Trajectory (FY2021–FY2025)
| Fiscal Year | Total Revenue | FFO/Share | Same-Store NOI Growth | Total IT Capacity (MW) |
|---|---|---|---|---|
| FY2021 | ~$4.3B | ~$6.53 | +1.5% | ~2,400 |
| FY2022 | ~$4.7B | ~$6.70 | +3.0% | ~2,800 |
| FY2023 | ~$5.5B | ~$6.55 | +3.5% | ~3,100 |
| FY2024 | ~$5.6B | ~$6.85 | +4.0% | ~3,300 |
| FY2025 | ~$5.85B | ~$7.00 | +4-5% | ~3,400 |
The pattern of FFO/share recovery through FY2024-FY2025 from the FY2023 transition period reflects the AI infrastructure demand cycle benefits combined with the operational integration of selected acquired capacity. Total IT capacity growing from approximately 2,400 MW in FY2021 to approximately 3,400 MW in FY2025 (approximately 42% cumulative capacity growth) reflects both organic development + selected portfolio expansion supporting the AI cycle deployment.
Hyperscaler Bookings Activity
| Period | Total Bookings ($M) | Hyperscaler Bookings % of Total |
|---|---|---|
| FY2022 | ~$520 | ~60% |
| FY2023 | ~$580 | ~70% |
| FY2024 | ~$1,050+ (record) | ~80%+ |
| FY2025 | ~$1,000-1,200 | ~80%+ |
The bookings (the metric tracking new leasing commitments that will translate into future revenue as deployments commission) growth from approximately $500M in FY2022 to approximately $1B+ in FY2024-FY2025 reflects the AI infrastructure demand cycle. The hyperscaler concentration of approximately 80%+ of bookings indicates that the AI demand surge is concentrated at hyperscaler customers (versus traditional enterprise data center customers where AI workload deployment has been more limited).
Development Pipeline + Capital Deployment
| Year | Development Capex ($B) | Stabilized Yield on Development |
|---|---|---|
| FY2022 | ~$2.5 | ~9-10% |
| FY2023 | ~$3.0 | ~8-9% |
| FY2024 | ~$3.5 | ~8-9% |
| FY2025 | ~$4.0+ | ~8-9% |
Digital Realty's development capex elevated to approximately $4B+ in FY2025 reflects the AI demand cycle development response. The development yield of approximately 8-9% on cost (the development of new data center capacity at construction cost yielding rental rates that exceed cost-to-yield benchmarks) supports value creation versus acquired properties at market valuations — though the elevated capex requires substantial financing through debt + selected equity issuance.
Capital Structure + Dividend
| Year | Net Debt ($B) | Dividend/Share | Dividend Coverage (AFFO) |
|---|---|---|---|
| FY2022 | ~$15.5 | $4.88 | ~95% |
| FY2023 | ~$18.0 | $4.88 | ~98% |
| FY2024 | ~$17.0 | $4.88 | ~95% |
| FY2025 | ~$18.0 | $4.88 | ~93% |
Digital Realty maintained the $4.88/share dividend through the multi-year period despite tight dividend coverage at approximately 93-98% of AFFO — reflecting management commitment to dividend continuity supporting income-oriented investor demand.
Market Evaluation
Digital Realty Trust trades at approximately 22-28x forward FFO/share — premium data center REIT multiples that reflect the AI infrastructure cycle exposure premium. The bull case is AI demand continuation + development pipeline + selected emerging strategic initiatives: if AI infrastructure capex continues at elevated levels through FY2027 supporting continued hyperscaler bookings + lease activity, if Digital Realty's development pipeline ($3-5B annual capex) translates into operating capacity at attractive yields, and if same-store NOI growth sustains at 4-5%, FFO/share could approach $7.50-8.50 by FY2027 with sustained multiple. The bear case is AI cycle reset + dividend coverage pressure + competitive intensity from Equinix + selected emerging: if AI hyperscaler capex normalizes following the FY2024-FY2025 surge, if Digital Realty dividend coverage continues compressing requiring selected dividend cut, or if Equinix competitive intensity at hyperscaler customers escalates, FFO/share growth could moderate with multiple compression.
The AI Infrastructure Cycle and Data Center Wholesale Capacity Strategic Position
The strategic argument that defines Digital Realty Trust's contemporary investment thesis rests substantially on the AI infrastructure capital expenditure cycle that has driven elevated hyperscaler data center demand through FY2023-FY2025. The strategic insight: AI accelerator chips (NVIDIA H100/H200/B100/B200, AMD MI300, Google TPU, plus emerging custom AI silicon at hyperscalers) require substantially more data center capacity per unit of compute than traditional CPU-based workloads — single AI training cluster requires 100-300+MW of data center capacity, and the hyperscaler AI infrastructure deployments at this scale support multi-year wholesale data center demand.
Digital Realty's hyperscaler positioning: Digital Realty has been the second-largest US wholesale data center provider (behind Equinix's combined retail + wholesale operations) for hyperscaler customer relationships. The hyperscaler customer concentration: Microsoft + Google + Amazon + Meta + Oracle + selected emerging hyperscaler-class customers collectively represent approximately 60%+ of Digital Realty's revenue. The customer concentration creates both opportunity (hyperscaler AI capex translates directly into Digital Realty revenue + leasing activity) + risk (customer concentration creates exposure to selected hyperscaler decisions affecting Digital Realty bookings + capacity utilization).
The development cycle response: Digital Realty's development capex elevated to approximately $4B+ in FY2025 reflects the strategic decision to expand capacity supporting the AI demand cycle. The development cycle requires multi-year capital deployment (data center construction takes 18-24 months from groundbreaking to commissioning) plus selected emerging strategic capabilities (power infrastructure development, cooling infrastructure for AI-optimized data centers requiring liquid cooling for high-density GPU deployments, plus selected emerging technical capabilities).
The competitive context: Equinix (covered separately as the largest data center REIT combining retail + wholesale operations) competes with Digital Realty at hyperscaler accounts plus has structural advantages in retail data center (Equinix's interconnection density supports retail enterprise customers requiring connectivity to multiple cloud + network providers). Selected smaller emerging data center operators (QTS Realty Trust acquired by Blackstone 2021 + selected emerging private equity-owned operators + selected hyperscaler-developed self-built data center capacity) compete at hyperscaler accounts. Digital Realty's competitive defense rests on multi-decade customer relationships + global operational scale + selected emerging technical capabilities at AI-optimized data centers.
The dividend coverage dynamic represents the structural concern affecting Digital Realty's investment narrative. The $4.88/share dividend (~3-4% yield) consumes approximately 93-98% of AFFO supporting the operational cash flow generation but with limited coverage cushion for selected adverse scenarios. The FY2026-FY2027 trajectory depends on whether AFFO growth supports continued dividend stability or whether selected operational pressures force selected dividend reduction (which would compromise Digital Realty's high-yield REIT positioning).
