For information only. Not investment advice, a recommendation, an offer or a solicitation, and without regard to any recipient’s objectives or circumstances. Figures come from third-party sources believed reliable but not guaranteed, and market data may be delayed. Past performance is not a reliable indicator of future results. |
AI Landlords Data Center REITsSummaryAI is creating an unprecedented wave of demand for data center capacity. AI workloads projected to grow from 9% of data center power consumption to 37% by 2030. This fuels a $4T-$5T capex cycle through 2030, focused on building a new fleet of power-intensive hyperscale facilities 25x larger than the average data center. Key Statistics
The Structural ThesisDemand Driver: AI InferenceThe shift from AI training to inference creates a structural tailwind for geographically distributed colocation networks. Inference requires <20ms latency, necessitating facilities near population centers. The Moat: Land & Power
Pricing Power
Company ImplicationsDigital Realty Trust (DLR | $174)
Equinix (EQIX | $1,023)
Risks
RelatedExposed companies (1) Vertiv Holdings CoVRT US |
Research Theme
AI Landlords Data Center REITs
Exposed Universe1 Companies
Last Updated2026-07-14
Loading graph…
Summary
AI is creating an unprecedented wave of demand for data center capacity. AI workloads projected to grow from 9% of data center power consumption to 37% by 2030. This fuels a $4T-$5T capex cycle through 2030, focused on building a new fleet of power-intensive hyperscale facilities 25x larger than the average data center.
Key Statistics
- AI Power Consumption Growth: 9% to 37% of total data center GW by 2030
- Capex Cycle: $4T-$5T over 2025-2030 (McKinsey: $3.8T-$7.9T)
- Capex Breakdown: 64% IT hardware (GPUs, CPUs, SSDs), 30% data center infrastructure (cooling, power), 6% power (grid connections)
- REIT Development Pipelines: Up 55% Q/Q to $18B in Q1 2026
- Data Center Development: Now 40% of all REIT development spending (up from 12% two years ago)
- Average Hyperscale Facility: ~1,000 MW (vs 40 MW average data center)
The Structural Thesis
Demand Driver: AI Inference
The shift from AI training to inference creates a structural tailwind for geographically distributed colocation networks. Inference requires <20ms latency, necessitating facilities near population centers.
The Moat: Land & Power
- Power availability is the defining barrier to entry. Northern Virginia's Dominion Energy has paused new connections.
- Incumbents' cost advantage: Long-duration PPAs signed at 2021-2024 rates are below current market rates.
- EQIX (2 GW) and DLR (1.2 GW) already have large under-construction pipelines.
Pricing Power
- Cash rental rate increases on leases >1 MW achieving double-digit growth.
- Record Q1 2026 bookings: DLR $707M, EQIX $378M.
Company Implications
Digital Realty Trust (DLR | $174)
- 310 data centers across 43.2M rentable sq ft.
- Q1 2026: Revenue $1.64B (+16.2% Y/Y), Core FFO $2.04.
- Pipeline: $16.5B with 1.2 GW under construction.
- Quant profile: strong Momentum and Growth offset by a very expensive Value reading.
Equinix (EQIX | $1,023)
- 281 data centers across 77 markets, 5 continents.
- Q1 2026: Revenue $2.44B (+10% Y/Y). Adj. EBITDA margins 51% (record).
- Quant profile: near-top Momentum, Quality and Growth, with Value the single weak component.
Risks
- Power Availability: Aging utility grids overwhelmed.
- Supply Chain: Lead times for high-voltage transformers.
- Tenant Concentration: Hyperscalers are both customers AND competitors.
- Interest Rate Sensitivity: REITs rely on external funding.