Over the past two decades, every real estate cycle has had its defining asset class such as retail in the early 2000s, multifamily through the recovery years, and industrial logistics during the e-commerce boom of the last decade. In the current times, a new contender is emerging as the center of investor attention: data centers.
They were once considered a niche, technically complex corner of the market, but have now rapidly become one of the most sought-after investment categories in commercial real estate.
The driving force behind this shift is the widespread AI workloads, enterprise cloud migration, and the broader digitization of nearly every industry. This is creating a sustained, long-term demand for compute capacity, and the existing infrastructure simply can’t keep pace with such heavy demand. Unlike past real estate booms that were driven by population growth or consumer spending patterns, this one is driven by the underlying architecture of the digital economy itself.
Data Centers are The New Industrial
Why Do We Say That?
Industrial real estate was significant in the last decade. Data centers have taken the reins now. AI adoption, cloud migration, and edge computing are creating a structural demand surge, not a passing trend. Vacancy in Tier-1 markets is near zero, and absorption is running ahead of new supply. Capital investment in U.S. data centers summed up to around $387 billion in 2025, and is expected to rally by about 81% to $700 billion in 2026.
Why Compare with Industrial?
Four structural features make data centers underwrite like the best industrial deals:
- Long NNN leases: 10-20-year terms with rent escalations written into the contract from day one.
- Mission-critical infrastructure: data centers power essential business operations, making relocation and downtime extremely costly for tenants.
- High barriers to entry: limited power availability, complex development requirements, and high replacement costs keep new supply scarce.
- Long-term growth orientation: demand is driven by secular trends (cloud computing, AI, digitization) not short-term economic cycles.
What is Different?
- Industrial is measured in square feet, Data centers are measured in megawatts (MW).
- Industrial values highway access and labor availability, Data centers value power capacity, fiber connectivity, and low latency.
- Industrial requires standard building capex, Data centers require significant investment in power, cooling, and redundant infrastructure.
How We Underwrite: Four Perspectives
The phrase “how we underwrite a data center deal” means something different depending on where you sit in the value chain. Let us discuss the 4 different approaches:
The Ground-Up Developer (Entitle, Build, and Stabilize)
The ground-up developer takes a project from raw land all the way through entitlement, construction, and stabilization. Returns depend on securing power and approvals efficiently, controlling development costs, and delivering a facility that attracts long-term tenants or buyers.
- Site & Power: Verify that utility-backed power capacity can realistically be secured and understand the full timeline involved. Substation queue positions, transmission upgrade costs, water availability for cooling, and fiber connectivity must be addressed before construction can be financed and remain contractually secure enough to support long-term operations.
- Lease Structure: Whether the project is pre-leased or speculative, the key question is whether the site’s power and connectivity profile is strong. For leased projects, evaluate rent escalations, lease term, and renewal options to ensure the asset can support financing and maintain long-term value.
- Income Analysis: Balance total development costs (including land acquisition, entitlements, permitting, generators, switchgear, and cooling infrastructure) against projected stabilized rents or comparable site values to determine development margins.
- Exit Assumptions: Assess entitlement and power agreement timelines, stabilized cap rates, lease rollover risk, and the property’s ability to be sold or refinanced. Also consider whether permitting delays, grid constraints, or construction setbacks could compress returns before exit.
The Converter (Acquire or Lease an Existing Structure and Retrofit It)
The converter creates value by transforming an existing building into a data center rather than developing one from the ground up. The strategy is built around saving time, reducing permitting risk, and accelerating revenue generation.
- Site & Power: Evaluate whether the building’s existing electrical infrastructure and utility connection can be upgraded to support data center operations. Compare the cost and timeline of these upgrades against a new development.
- Lease/Acquisition Structure: Analyze the cost of acquiring or leasing the property and completing the retrofit. The investment thesis depends on achieving faster delivery and lower development risk than a ground-up project.
- Income Analysis: Compare retrofit costs with projected stabilized rents. The key advantage is reaching revenue generation sooner than a newly constructed facility.
- Exit Assumptions: Assess whether the converted asset can be sold or refinanced as a competitive data center. Consider structural limitations, future expansion potential, and any unforeseen retrofit costs that may affect value.
The Master Lessee (Lease the Whole Site, Sublease in Pieces)
The master lessee leases a large block of data center capacity and generates value by subleasing it to multiple tenants. Success depends on leasing space faster and at higher rates than the cost of the master lease.
- Site & Power: Determine whether the site’s power and cooling capacity can be efficiently divided and leased to multiple users with varying requirements.
- Lease Structure: Analyze the spread between master lease costs and expected sublease revenue, while accounting for occupancy ramp-up and leasing restrictions.
- Income Analysis: Model tenant absorption, occupancy growth, blended rental rates, and vacancy risk during the lease-up period.
- Exit Assumptions: Consider lease expiration, assignment rights, renewal risk, and whether the position can be sold before the master lease ends.
The End User (Lease for Own Use)
The end user leases data center space to support its own operations rather than to generate rental income. The decision is driven by operational efficiency, reliability, and long-term cost considerations.
- Site & Power: Confirm that power density, redundancy levels (N+1 or 2N), and network connectivity meet the technical requirements of the workloads being deployed.
- Lease Structure: Compare the cost and flexibility of leasing against owning, with particular attention to expansion rights and future capacity needs.
- Income Analysis: Focus on total occupancy costs versus the capital expenditure, depreciation, and operational burden of owning a facility.
- Exit Assumptions: Evaluate renewal, expansion, and relocation risks, including downtime, migration costs, and potential business disruption.
What Are Some Red Flags To Watch Out For?
Irrespective of the side you’re investing from, some red flags can stop a deal before underwriting even begins:
- Power Constraints and Grid Delays: If there isn’t enough committed utility capacity, a data center project can quickly become unfinanceable.
- Single-Tenant Concentration: When more than 80% of revenue comes from one hyperscale tenant, cash flow and lease rollover risks increase manifold.
- Aging Infrastructure: A Power Usage Effectiveness (PUE) above 1.6 often implies outdated systems, higher operating costs, and future capex requirements.
- Permitting Delays: Slow approvals can delay project completion, push back stabilization, and reduce returns for investors.
- Local Opposition: Community resistance about resource exhaustion and restrictive regulations can delay or even stop projects, which can make it a risky venture regardless of strong market fundamentals.
- Rapidly Changing Tech: The AI industry is changing very quickly, which raises the question whether the current hardware would be compatible with how it develops. This creates a risk wherein the tenant might move to the data center facility that provides efficient hardware, or incur additional costs of changing the systems to retain the tenant and to adapt to the developments.
The investment case for data centers is one to note. The global data center market is expected to reach $400 billion by 2030, while it boasts of stabilized cap rates in the range between 4.5% and 6.5%. According to Goldman Sachs Research, AI workload demand is projected to triple by 2027. With data center developments typically taking around 18 months from site selection to delivery, investors who identify opportunities early and underwrite them correctly stand to benefit from significant forward-pricing advantages. At RealVal, we provide institutional-grade underwriting for digital infrastructure assets.
Book your consultation at info@therealval.com.
