Module XXIV·Article VI·~6 min read
Infrastructure: Data Centers and Tower Companies
Direct Real Estate Investment
Turn this article into a podcast
Pick voices, format, length — AI generates the audio
Digital Infrastructure: Data Centers and Telecom Towers
Digital infrastructure is the fastest-growing segment of alternative real estate investments, transformed by the explosive growth of data, cloud computing, 5G, and artificial intelligence. These assets combine real estate characteristics (long-term contracts, stable cash flows) with exposure to the technology sector.
Data Centers: Demand Drivers
Macro Drivers
- Cloud computing — migration of enterprise IT to the cloud (AWS, Azure, GCP)
- AI/ML workloads — generative AI requires massive compute capacity
- 5G deployment — edge computing for low latency
- IoT proliferation — billions of connected devices
- Video streaming — Netflix, YouTube, TikTok generate massive traffic
- Gaming — cloud gaming and VR/AR
Quantitative Indicators
- Global data volume doubles every 2-3 years
- AI compute demand grows 10x annually (2023-2025)
- Hyperscale data center capacity grew from 20% to 60% of the global market in 10 years
- Global data center market: ~$250 billion (2024), forecasted to $500+ billion by 2030
Typology of Data Centers
-
Hyperscale Data Centers
Owners: Big Tech (Amazon, Microsoft, Google, Meta)
Size: 100+ MW, hundreds of thousands of sq ft
Characteristics: Single-tenant, custom-built
Lease term: 10-20 years
Investment opportunity: Build-to-suit development, powered shell leases -
Colocation (Colo) Data Centers
Model: Multi-tenant facilities, retail and wholesale
Contract size: From 1 rack to 1+ MW
Services: Power, cooling, connectivity, cross-connects
Lease term: 3-7 years (retail), 5-10 years (wholesale)
Key operators: Equinix, Digital Realty, CyrusOne, CoreSite -
Edge Data Centers
Purpose: Low-latency applications, close to end users
Size: 1-5 MW
Applications: Gaming, autonomous vehicles, IoT, CDN
Locations: Tier 2/3 cities, network aggregation points
Hyperscale vs Colocation: Comparison
| Characteristic | Hyperscale | Colocation |
|---|---|---|
| Tenant concentration | Single tenant | Multi-tenant |
| Credit quality | AAA (Big Tech) | Mixed (IG to HY) |
| Lease length | 10-20 years | 3-10 years |
| Capex intensity | Tenant builds out | Landlord provides infra |
| Pricing | $/kW/month | $/kW/month + connectivity |
| Development yield | 8-12% | 10-15% |
| Stabilized cap rate | 5-7% | 5-6% |
Data Center Economics
Key Metrics
- MW (Megawatts) — primary capacity measurement unit
- PUE (Power Usage Effectiveness) — energy use efficiency (target: lt; 1.4$)
- Uptime SLA — uptime guarantee (99.99% = 52 minutes downtime/year)
- $/kW/month — rent rate per kilowatt per month
- Cross-connects — connections between clients (additional revenue)
Typical Cost Structure
| Category | Share of total cost |
|---|---|
| Land and site | 5–10% |
| Shell building | 15–20% |
| Power infrastructure | 30–35% |
| Cooling systems | 15–20% |
| IT fit-out (tenant or landlord) | 20–25% |
AI/GPU Impact
Generative AI radically changes data center requirements:
- Power density: AI racks require 50–100+ kW vs 5–10 kW for standard racks
- Cooling requirements: Liquid cooling becomes a necessity
- Power availability: Critical bottleneck in key markets
- Premium pricing: AI-ready capacity receives significant premium
Tower Companies: Business Model
What is a tower company?
Tower companies own and lease telecommunications infrastructure — towers, rooftops, small cells — to mobile network operators (MNOs: AT&T, Verizon, T-Mobile, etc.).
Business Model
- Master Lease Agreements (MLAs) — long-term contracts with MNOs
- Colocation — multiple operators on one tower
- Amendment revenue — upgrade existing equipment
- Escalators — built-in annual rental growth (3% per year)
- Churn — minimal (2–3% annually)
Types of Tower Assets
- Macro towers — traditional towers 50–300 ft high
- Rooftops — equipment on building rooftops
- Small cells — low-power nodes for densification
- DAS (Distributed Antenna Systems) — inside buildings and stadiums
- Fiber — fiber optic networks (vertical integration)
Key Players
| Company | Towers (global) | Market Cap | Geographic Focus |
|---|---|---|---|
| American Tower (AMT) | ~225,000 | ~$85 bn | US, LatAm, EMEA, India |
| Crown Castle (CCI) | ~40,000 + fiber | ~$45 bn | US only |
| SBA Communications | ~39,000 | ~$22 bn | US, LatAm |
| Cellnex | ~130,000 | ~$25 bn | Europe |
Tower Economics
Key financial characteristics:
- Recurring revenue: 98%+ of total revenue
- EBITDA margin: 60–65%
- Capex intensity: Low after initial build
- FCF conversion: High
- Dividend yield: 3–5% (REIT structure)
Tenancy and Co-location
- Average tenants per tower: 1.5–2.5 in developed markets
- Incremental margin: 90%+ for each additional tenant
- Revenue per tower: $25,000–50,000/year in the US
5G Impact
5G deployment is a major growth driver:
- More spectrum: Requires equipment upgrades on towers (amendment revenue)
- Densification: More small cells in urban areas
- New sites: Coverage expansion
- ORAN: Open RAN might change competitive dynamics
Infrastructure REITs
Digital infrastructure is primarily traded via REITs (Real Estate Investment Trusts):
| REIT | Focus | FFO/Share Growth | Dividend Yield |
|---|---|---|---|
| Equinix (EQIX) | Colocation DC | 8–10% | 2.0% |
| Digital Realty (DLR) | Hyperscale + Colo | 6–8% | 3.5% |
| American Tower (AMT) | Towers global | 8–10% | 3.0% |
| Crown Castle (CCI) | Towers + Fiber (US) | 5–7% | 5.5% |
Renewable Energy Infrastructure
Renewable energy infrastructure is an adjacent segment:
Asset Types
- Solar farms — utility-scale and distributed
- Wind farms — onshore and offshore
- Battery storage — grid-scale energy storage
- EV charging infrastructure — charging networks
- Green hydrogen — production facilities
Investment Characteristics
- Long-term PPAs: Power Purchase Agreements for 15–25 years
- Contracted revenue: Predictable cash flows
- Government support: Tax credits, subsidies (IRA in the US)
- Inflation-linked: Many PPAs are inflation-indexed
- Commodity exposure: Merchant risk in the absence of PPA
Digital Infrastructure Trends
Key trends for 2024–2030:
- AI-driven demand surge: GPU clusters require massive DC capacity
- Power scarcity: Availability of power is the main bottleneck
- Sustainability focus: Carbon neutrality, renewable energy procurement
- Edge expansion: Decentralization of compute for low latency
- Hyperscaler insourcing: Big Tech is building more own DCs
- Tower consolidation: M&A activity in tower segment
- 5G Advanced and 6G: Next-gen wireless demands infrastructure upgrade
Key Metrics for Evaluation
| Metric | Data Centers | Towers |
|---|---|---|
| Primary metric | $/kW/month | Revenue per tower |
| Capacity utilization | % of IT load deployed | Tenants per tower |
| Growth metric | MW under construction | New builds + amendments |
| Profitability | EBITDA margin (50–60%) | EBITDA margin (60–65%) |
| Valuation | EV/EBITDA (15–25x) | EV/EBITDA (18–25x) |
| Development yield | 10–15% | Build: 8–12%, Acquire: 5–7% |
Recommendations for CIOs
- Secular growth exposure — digital infra is a beneficiary of multiple tech trends
- Quality operators — sector dominated by well-capitalized players
- AI thematic — overweight DCs with AI-ready capacity
- Public vs Private — REITs offer liquidity, private offers better returns
- Power access — critical differentiator for DC investments
- Geographic diversification — mix US and international exposure
- Development vs stabilized — development offers higher returns but execution risk
§ Act · what next