Data center CEO REVEALS MAJOR threat to America's AI BOOM

Digital Realty CEO Andy Power discussed an AI-driven data-center expansion, emphasizing rising demand for digital infrastructure. The discussion also addressed community concerns about the sector's electricity and water consumption, grid capacity, and broader infrastructure requirements. No financial results, guidance, or quantified operating impacts were disclosed.

Analysis

The investable issue is not incremental AI leasing demand—already embedded in data-center REIT valuations—but whether DLR can convert contracted capacity into returns above its rising cost of capital. Grid interconnection delays and customer-funded power infrastructure can defer rent commencement while construction interest and development spending accrue, creating a near-term FFO-per-share versus backlog conversion gap. Markets will reward disclosed signed MW, but should distinguish between leases with secured utility service dates and capacity contingent on transmission upgrades.

DLR's most material second-order exposure is local power-market scarcity. Utilities and merchant generators serving Northern Virginia, Dallas, Chicago and secondary AI campuses gain pricing and load-growth visibility, while power-constrained markets can shift hyperscaler buildouts toward regions with available generation rather than toward the lowest-latency locations. Community opposition is principally a permitting-duration and capex-risk issue, not an immediate demand impairment; it becomes financially relevant if mandated backup-power, water, transmission, or grid-contribution requirements push development yields below DLR's stabilized asset yields.

Consensus may be underestimating the customer-concentration bargaining effect. Hyperscalers need capacity, but their scale permits them to demand phased delivery, termination rights and pass-through structures that limit DLR's upside if electricity prices rise. Over the next 1-3 months, the key catalyst is management disclosure on pre-leasing, secured power, development yield and funding mix; over 6-18 months, the thesis depends on whether AI demand translates into sustained absorption rather than a concentrated wave of speculative reservations.

Key Decisions for Investors

• Maintain DLR as a watch-list long rather than chase the AI narrative; initiate only after earnings confirm leased capacity with utility-secured energization dates and development yields meaningfully above the company’s marginal funding cost. Thesis is invalidated by a cut to core FFO guidance, material project delays, or rising customer concessions.
• For a relative AI-infrastructure expression, consider long DLR / short VNQ over a 6-12 month horizon only if DLR demonstrates backlog conversion; this isolates data-center demand from broad REIT rate sensitivity. Stop the spread if 10-year Treasury yields rise sharply without corresponding DLR leasing or pricing revisions.
• Monitor regulated utilities with direct exposure to data-center load growth in constrained service territories as a potentially cleaner second-order beneficiary than DLR. Do not initiate from this item alone: require utility-specific evidence that incremental load earns authorized returns without adverse regulatory cost-allocation rulings.
• Set an event alert around DLR quarterly disclosures for signed versus commenced MW, power-secured pipeline, tenant concentration and capex per MW. A widening gap between signed and revenue-producing capacity is a signal to avoid or hedge long exposure through the next 1-3 quarters.