Digital Realty at Bank of America conference: ai demand lifts outlook

Digital Realty raised its outlook to double-digit FFO growth for 2025 and beyond, supported by $2.3 billion of signed-but-not-commenced leases—more than 30% of its roughly $5 billion data-center revenue base. Its $20 billion development pipeline carries an expected 11.5% initial cash yield, while record $108 million quarterly zero-to-one/interconnection leasing and 25% overall re-leasing spreads underscore AI-driven demand and tightening capacity. The company has raised $3.25 billion of LP equity for its first closed-end fund and maintains 4.7x debt-to-EBITDA, though labor shortages, constrained power/site availability and local regulatory opposition could slow delivery.

Analysis

DLR’s earnings power is shifting from a rate-sensitive REIT model toward a contracted infrastructure-development model, which should support a lower FFO-risk premium if lease commencements convert on schedule. The key incremental upside is not another exceptional quarterly renewal spread, but the combination of embedded rent resets and a capital-light fee stream from third-party vehicles; that mix can lift growth without requiring equivalent common-equity issuance. The constraint is execution: a delay in energized delivery converts signed demand from a revenue catalyst into a working-capital and construction-cost burden.

Power and skilled-trade scarcity should widen the gap between scaled operators with pre-secured sites, procurement capacity, and investment-grade balance sheets versus smaller private developers dependent on project financing. This is modestly positive for DLR and, indirectly, BX through demand for institutional data-center capital, but it raises the probability that supply shifts to secondary power markets rather than disappearing. Utilities and independent power producers in new data-center corridors—not merely Northern Virginia—are likely to capture more durable economics than landlords if interconnection queues become the binding constraint.

Consensus is likely extrapolating both AI leasing and elevated renewal pricing too linearly. The near-term risk is not weak demand; it is tenant concentration, commissioning slippage, and a higher-for-longer cost of capital compressing the value of development cash flows. Over the next 1-3 months, monitor disclosed commencements, construction spend versus budget, and new fund commitments. Over 6-18 months, local permitting or grid-upgrade requirements could lower the growth multiple even if contracted revenue remains intact; a material reduction in development yields, backlog conversion, or leasing spreads would falsify the bullish case.

Key Decisions for Investors

• Initiate a modest long DLR position only on evidence of on-time commencements in the next earnings release; target a 12-18 month hold. Underwrite upside from FFO estimate revisions and fee-related earnings, but cap risk if construction costs rise enough to reduce expected stabilized development yields below 10% or leverage trends above 5.0x.
• Express relative value via long DLR / short a broad office REIT proxy such as BXP over 6-12 months. The trade isolates scarce-power, contracted digital infrastructure from structurally challenged urban office supply; reassess if Treasury yields fall sharply enough to create a broad office multiple catch-up.
• Add BX on weakness rather than chase DLR’s move: institutional demand for data-center exposure should support fundraising and asset-management fees over 12-24 months, while BX’s diversified platform reduces single-project execution risk. Falsifier: weaker private-market fundraising or evidence that data-center development returns are being competed down.
• Do not add directional NVDA exposure from this signal alone. Delivery bottlenecks can defer customer deployment schedules even while long-run compute demand remains intact; watch hyperscaler capex guidance and power-ready capacity before treating data-center leasing strength as incremental GPU revenue acceleration.