Data Center & Colocation Feasibility Study

Power, connectivity, cooling, and tenant demand analysis for SBA-financed data center and colocation projects

Data centers represent one of the most capital-intensive and technically complex property types in SBA lending. A purpose-built data center with raised floors, redundant power distribution, precision cooling systems, fire suppression (clean agent), and physical security infrastructure is unambiguously a special purpose property. The facility cannot be converted to standard commercial use without demolition-scale renovation. Under SOP 50 10 8, this classification triggers a feasibility study requirement and, under the 504 program, increases equity injection from 10% to 15% (20% for startups).

The data center market is experiencing unprecedented demand driven by AI workloads, cloud computing expansion, and digital transformation. Total U.S. data center capacity exceeded 5,000 MW in 2025, with demand outpacing supply in most primary and secondary markets. This supply-demand imbalance creates opportunities for SBA-financed edge, colocation, and small-scale enterprise data centers that serve the growing segment of the market too small for hyperscale operators but too large for on-premise server closets.

Data Center Models and SBA Fit

Not all data centers are viable SBA projects. The feasibility study must assess which model the borrower is pursuing and whether it fits within SBA lending parameters.

The SBA-viable segment is retail colocation, edge, enterprise, and modular facilities in the $1M to $5M project cost range. These are the facilities the feasibility study is designed to evaluate. Hyperscale and wholesale colocation projects exceed SBA lending limits and use institutional capital markets for financing.

Power Availability and Cost Analysis

Power is the single most important input in data center feasibility. It determines the facility's capacity ceiling, its largest ongoing operating cost, and its competitiveness relative to other markets. The feasibility study evaluates power at three levels.

Grid Capacity and Utility Infrastructure

The local utility must have sufficient grid capacity to serve the facility's power requirements. A 1 MW data center draws approximately the same power as 750 homes. In some markets (Northern Virginia, parts of the Pacific Northwest, central Ohio), utility capacity is constrained, with new service requests taking 12 to 36 months to fulfill. The feasibility study confirms available utility capacity at the subject site, estimates the interconnection timeline, and identifies any infrastructure upgrade costs (transformer installation, line extensions) that the borrower must fund.

Electricity Pricing

Electricity cost varies from $0.04/kWh in low-cost markets (parts of the Pacific Northwest, Texas, Oklahoma) to $0.15+/kWh in high-cost markets (California, New England, Hawaii). For a 1 MW facility operating at 80% load, annual electricity cost ranges from approximately $280,000 at $0.04/kWh to over $1 million at $0.15/kWh. This 3.5x cost differential makes site selection a strategic decision that the feasibility study must quantify.

Redundancy and Backup Power

Data center power systems are classified by tier (Uptime Institute Tier I through Tier IV), with each tier specifying redundancy requirements. SBA-financed facilities typically target Tier II (redundant capacity components) or Tier III (concurrently maintainable), which require UPS systems ($100,000 to $500,000 for 1 MW), diesel generators ($200,000 to $600,000 per MW), automatic transfer switches, and fuel storage. These infrastructure costs are significant line items in the capital budget that the feasibility study must include.

Connectivity and Fiber Infrastructure

A data center without network connectivity is a warehouse with expensive cooling. The feasibility study assesses the site's fiber infrastructure and network ecosystem, which directly determines the facility's attractiveness to tenants.

Carrier diversity. Colocation tenants require access to multiple network carriers for redundancy. A site served by a single fiber provider is less competitive than one with 3 to 5 carrier options. The feasibility study inventories the lit fiber providers serving the site or available within a reasonable construction distance (typically under 1 mile for new lateral builds, which cost $50,000 to $200,000+ depending on distance and terrain).

Latency requirements. Edge data centers and low-latency applications (financial trading, gaming, real-time AI inference) require sub-5ms roundtrip latency to end users or cloud on-ramps. The feasibility study maps the site's latency to major cloud regions (AWS, Azure, Google Cloud) and population centers to validate that the location meets the target applications' performance requirements.

Internet exchange proximity. Proximity to internet exchange points (IXPs) and cloud on-ramps adds value for colocation facilities. The study identifies the nearest IXPs and assesses whether the site can offer competitive network performance relative to established colocation markets.

Cooling Efficiency: PUE Benchmarking

Power Usage Effectiveness (PUE) measures how efficiently a data center uses energy. It is calculated as total facility power divided by IT equipment power. A PUE of 1.0 is theoretically perfect (all power goes to computing); real-world facilities range from 1.1 (best-in-class hyperscale) to 2.0+ (older, inefficient facilities).

The difference between a 1.3 PUE and a 1.8 PUE on a 1 MW facility is approximately $350,000 per year in additional power cost. This directly impacts operating margin and DSCR. The feasibility study models the proposed facility's PUE based on the cooling design and climate zone, then projects total power cost (IT load plus cooling overhead) as the primary operating expense. New construction should target PUE below 1.4 to remain competitive with modern colocation facilities.

Tenant Demand Analysis by Market Tier

Data center markets are classified by tier, which reflects the depth of demand, supply, and network infrastructure. The feasibility study assesses demand within the context of the market tier.

SBA-financed data centers are most viable in Tier 2 and Tier 3 markets where institutional capital has not yet saturated the supply, land and power costs are lower, and local enterprises need colocation services but lack access to Tier 1 facilities. The feasibility study identifies the specific demand drivers in the target market: healthcare organizations with HIPAA hosting needs, financial firms requiring local disaster recovery, government agencies with data sovereignty requirements, and managed service providers seeking wholesale colocation for their client base.

Construction Cost Per MW

Data center construction cost varies dramatically by design, redundancy tier, and whether the facility is a new build, shell conversion, or modular deployment.

For SBA-financed projects, shell conversions and modular deployments offer the best cost-to-revenue ratio. Converting an existing warehouse or industrial building with adequate power service ($3M to $6M per MW) produces a facility at 40% to 60% of the cost of a purpose-built Tier III data center. The feasibility study evaluates the conversion candidate's structural capacity (floor loading for server racks: minimum 150 lbs/SF), ceiling height (minimum 10 feet clear for hot/cold aisle containment), and electrical service capacity.

SBA Eligibility for Data Center Projects

Owner-operator model. The SBA requires that the borrower's business occupy the property. For a colocation facility, the borrower is the data center operator providing colocation services to tenants. This is analogous to a self-storage operator or apartment building owner: the borrower operates the facility business, not just the real estate. Both 7(a) and 504 are viable for this model.

Enterprise (single-tenant) model. A business building a data center for its own IT operations occupies the facility with its own business. This is the cleanest SBA eligibility path. The 504 owner-occupancy requirement (51% existing, 60% new construction) is easily met when the borrower's own servers fill the facility.

Special purpose classification. Data centers are treated as special purpose due to specialized infrastructure. Under 504, this means 15% equity (50/35/15) for established operators and 20% (50/30/20) for startups. On a $4 million project, the difference between 10% and 20% equity is $400,000, which is significant for SBA borrowers.

Equipment vs. real estate. Data center projects have a high equipment-to-real-estate ratio. Servers, networking, UPS systems, generators, cooling equipment, and fire suppression can exceed the real estate cost. The SBA loan can encompass both the real estate and the equipment (particularly under 504), but the feasibility study must clearly delineate the capital budget components and model debt service on the total project cost.

Cost and Timeline

Data center feasibility studies are among the most complex and typically range from $4,900 to $25,000 or more. Small-scale edge or enterprise facilities in markets with established data center ecosystems fall toward the lower end. Colocation facilities requiring detailed tenant demand analysis, multi-MW projects, or greenfield sites requiring utility capacity and fiber studies push toward the upper end.

Standard delivery runs 15 to 20 business days, longer than most property types due to the technical analysis required (power capacity verification, fiber infrastructure mapping, PUE modeling, tenant demand assessment). For detailed pricing across all property types, see our cost and timeline guide.