Power, interconnection timing, absorption and tenant credit — the four things that decide whether a data center project is financeable, tested against evidence rather than pipeline optimism.
Data center demand is not scarce; deliverable power is. A site's viability turns on how much capacity the utility can actually energise, when, and under what conditions — substation capacity, transmission constraints, the interconnection queue position, and the utility's own load-growth commitments to other customers in the same territory.
A feasibility study that treats a letter of intent from a utility as settled capacity is not underwriting the project. We work the power question first: documented available capacity, queue position and study status, the timeline the utility itself is quoting, on-site generation or bridging arrangements, and the water or cooling constraints that ride alongside. Where the energisation date does not support the lease-up schedule, that is the finding, and everything downstream of it changes.
Colocation and multi-tenant projects are frequently underwritten on an absorption pace borrowed from the market's headline shortage. That shortage is real in some power markets and entirely absent in others, and it does not transfer across submarkets simply because both are described as data center corridors.
We test absorption against what has actually leased in the same power market at comparable configurations, against the pipeline that will deliver before the subject does, and against the tenant type the sponsor can realistically sign given the facility's density, redundancy tier and connectivity. Where the deal rests on a single anchor tenant, we analyse the covenant behind it and show coverage with and without that tenant, because a lender's exposure is defined by the downside case.
A complete independent study with the market and power analysis evidenced, the revenue and cost build shown, and a fully linked financial model with no hard-coded outputs, so the lender's underwriter can run their own sensitivities. Where the project is seeking SBA, USDA or conventional financing, the report is scoped to the reviewing program's requirements from the first day rather than retrofitted afterwards.
The fee is fixed, quoted before work begins, and never contingent on the finding.
A data center feasibility study prepared for a lender or reviewing agency is not a marketing document. It is a structured credit analysis that answers whether the project can service its debt under realistic operating conditions and under stress. For this asset class, the scope extends well beyond a generic income-and-expense model because the cost structure, revenue mechanics and infrastructure dependencies are materially different from conventional commercial real estate or light industrial projects.
Demand analysis for a data center differs from conventional real estate market studies because the relevant demand signal is not population or employment — it is power-hungry digital workload growth, which is measured through a different set of sources and counting methods. Data center feasibility study consultants who understand this asset class build the market section from infrastructure evidence, not from general economic proxies.
On the supply side, the analyst inventories existing and planned competing facilities by reviewing utility interconnection queue filings, which publicly disclose the location, requested capacity and queue position of large power users in the relevant transmission zone. State and local building permit records, corporate real estate filings with the SEC, and data center trade association databases are cross-referenced to distinguish announced projects from those with committed power and executed construction contracts. Planned capacity that lacks a confirmed interconnection agreement is treated as contingent, not competitive.
On the demand side, the analyst examines regional carrier-neutral exchange points, fiber route density, existing colocation absorption trends drawn from published market reports, and anchor-tenant lease commitments already in hand. Hyperscale demand signals — such as publicly disclosed cloud infrastructure expansion plans — are noted but not counted as revenue unless a binding agreement exists. The result is a supply-demand balance expressed in megawatts of leasable critical load, not in square feet alone, because power capacity is the binding constraint in this asset class.
Every feasibility model has a small number of inputs that account for most of the variance in the coverage ratio. For data center projects, those inputs are different from those in hospitality or retail studies, and a competent analyst isolates them explicitly so a credit officer can see exactly where the risk is concentrated. The fully linked Excel workbook Wert-Berater delivers has no hardcoded values, so any reviewer can move these inputs independently and observe the effect on debt-service coverage without unlocking a protected cell.
The credit questions a lender raises for a data center project are more infrastructure-specific than for most commercial loan types, and the feasibility study must address them directly rather than by analogy to conventional real estate.
For SBA 7(a) and 504 engagements, the study is prepared to SOP 50 10 8. The coverage minimums are 1.15x operating and 1.00x global. For a data center, the SBA reviewer will scrutinize whether projected revenue rests on executed agreements or on assumed absorption, and whether the power supply is contractually secured. A study that counts unexecuted letters of intent as stabilized revenue will not survive underwriting review.
For USDA Business & Industry and REAP engagements, the study follows RD Staff Instruction 5001. Rural data center projects — often sited for lower land cost, cooler ambient temperatures or proximity to renewable generation — must demonstrate that the local labor market can support technical operations staffing and that the utility infrastructure commitment is documented, not merely planned.
For conventional lenders, the coverage standard is typically 1.20x, and the concentration question dominates: a facility with one or two tenants representing the majority of critical load will face scrutiny of tenant credit quality, lease term remaining and the financial consequence of a single non-renewal. The study addresses each of these directly, with an explicit statement of conditions that identifies what must remain true for the projections to hold.
Wert-Berater quotes a fixed fee within one business day of receiving a project description. The fee does not change based on the finding, and no portion of it is contingent on a positive determination. This structure is not a marketing position — it is the condition that makes the study credible to a lender or reviewing agency. A fee tied to loan approval creates an incentive that a credit officer will recognize and discount.
Standard delivery is 10 to 15 business days from receipt of a complete data room. For data center projects, a complete data room includes the executed or draft power service agreement, the interconnection queue filing or utility confirmation letter, executed or draft tenant leases or colocation agreements, construction cost documentation, the proposed debt terms and any existing environmental or zoning approvals. Incomplete data rooms extend the timeline; the engagement clock does not start until the data room is confirmed complete.
Rush delivery is available and quoted at the time of engagement. Every engagement is published to a secure client portal where the financial model remains live after delivery. When a lender requests a revised scenario — a different loan amount, a changed interest rate, a modified absorption schedule — the model recalculates without requiring a new engagement. The bound narrative report, ten-year pro forma, sensitivity tables and ratio analysis benchmarked against RMA and IBISWorld data are all delivered as a single coordinated package, not as separate documents assembled after the fact.
A data center site feasibility study treats power as a documented delivery chain, not a marketing claim attached to a parcel. The analysis begins with the proposed critical IT load, total facility load, redundancy design and phased ramp, then reconciles those requirements with the utility's service territory, available voltage, substation configuration, feeder or transmission source and stated energisation process. A general capacity letter may support early screening, but underwriting also needs to know whether capacity is reserved, what agreements and deposits remain outstanding, which studies are complete and what network work must occur before the meter can be energised.
Utility verification is assembled from written correspondence, service applications, study reports, queue records, draft or executed service agreements and direct clarification through the sponsor's authorised utility contact. The study records what the utility has committed, what remains indicative and which conditions could change cost or timing. It also reconciles the power request with other large loads under study in the same service area. Nameplate substation capacity is not assumed to be available to the subject because firm load obligations, contingency criteria, transformer configuration and transmission constraints may limit deliverable capacity.
Substation proximity matters, but distance alone does not determine the solution. Route control for a new line, road or rail crossings, easements, right-of-way acquisition, switching and protection equipment, metering, utility design standards and transmission upgrades can all govern cost and schedule. The development programme identifies responsibility for each item and aligns utility milestones with design, procurement, commissioning and tenant delivery obligations. Bridge generation is analysed only where it has a credible fuel, emissions, noise, permitting and operating plan; it is not treated as an automatic substitute for delayed grid service.
Capacity planning separates utility input, total facility demand and saleable critical load so that revenue is not applied to power consumed by cooling and supporting systems. The model reflects the design power-usage effectiveness assumption, redundancy architecture and realistic utilisation at each phase. Shell space, electrical rooms and cooling plant may be built ahead of revenue-producing capacity, while some utility and substation costs must be incurred before the first tenant is live. The sources and uses schedule therefore follows the physical sequence of campus development rather than spreading costs evenly across an assumed stabilisation period.
Phases are tested as individual financing and operating decisions. Each phase needs a defined power block, construction scope, commissioning path, tenant or sales trigger and source of capital. A later phase should not subsidise early debt service unless its funding and delivery are demonstrably available. The analysis also considers whether common infrastructure is appropriately allocated, whether an initial build can operate independently if expansion pauses, and whether land and utility rights preserve the planned ultimate campus.
Resilience features affect both capital cost and market fit. Dual feeds, generators, uninterruptible power supplies, energy storage and redundant cooling are evaluated against the service level promised to tenants, not added as generic premiums. Maintenance bypass, fuel logistics, testing, replacement cycles and environmental restrictions are part of the operating case. A design can be technically robust yet financially mismatched if its redundancy cost exceeds what the target customer segment will pay.
Connectivity diligence identifies the carriers and fibre routes that can physically serve the site, the points of entry, route diversity and the network locations to which latency is commercially important. Carrier presence in a metropolitan area does not prove that service reaches a particular parcel or that two quoted paths are physically diverse. Route maps, carrier letters, easement information and proposed extensions are reconciled to distinguish lit service from nearby backbone fibre and to identify common ducts, bridges or rights of way that could create a single point of failure.
Latency is tested against the proposed operating model. Hyperscale cloud capacity, enterprise disaster recovery, content delivery, financial workloads and edge applications do not share one threshold or one set of destination nodes. The data center market study therefore relates measured or carrier-supported latency to the customer segment and key exchange points rather than claiming that a broad regional average is adequate. Cross-connect availability, carrier competition, recurring transport cost and the time needed to construct new laterals also enter the site comparison.
For lender review, the relevant question is whether the connectivity package is available by tenant commencement and remains defensible through the lease term. Unexecuted carrier discussions are identified as such. The downside case considers extension delay, higher construction contributions, lack of route diversity and the loss of a key carrier, particularly when the revenue case depends on network-rich colocation rather than a dedicated hyperscale deployment.
Cooling strategy is evaluated with the server density, climate, reliability target and utility environment. Air-cooled, evaporative, chilled-water and liquid-cooling systems have different power, water, maintenance and capital implications. The study reconciles the selected concept with the mechanical design basis and the financial model, including seasonal performance and the effect of future higher-density deployments. A low power-usage assumption is not accepted if the corresponding cooling equipment and operating plan are absent from the development budget.
Where water is used, diligence addresses the source, service commitment, connection and meter capacity, consumption profile, discharge, treatment and any reuse plan. Drought exposure is considered through utility restrictions, allocation policy and the practical availability of alternative operating modes, without assuming that a permit or service letter insulates the project from future constraint. Wastewater capacity, blowdown disposal and water-quality requirements can be as important as supply. Relevant permits and agency decisions are reported by status, with legal and engineering conclusions reserved to the responsible professionals.
Air permits for standby generation, fuel storage, noise, stormwater, wetlands, flood conditions, contamination and backup-power testing can affect both entitlement and operations. The study connects these issues to schedule, capital cost and tenant service obligations. If the cooling or resilience plan depends on an approval not yet obtained, the report states that dependency and tests the consequence of delay or a more costly compliant alternative.
A viable site must accommodate buildings, substations, electrical yards, cooling plant, generators, fuel systems, security setbacks, fibre entries, stormwater facilities, construction laydown and future phases. Parcel size is therefore tested against a concept plan and known constraints rather than against building area alone. Topography, geotechnical conditions, flood exposure, access for heavy equipment, adjacent uses and route control for power and fibre can distinguish apparently similar candidate sites.
Zoning review identifies whether the proposed use is permitted, conditional or requires a legislative change, then traces site-plan, building, electrical, environmental and utility approvals. Data centres may prompt local questions about noise, water, generator emissions, visual impact, construction traffic, emergency response, employment, tax base and the allocation of scarce power. The study records hearing and notice requirements, development-agreement discussions and commitments already made. It does not predict political approval where the public process remains open.
Tax abatements, sales or use tax treatment, property-tax agreements, infrastructure participation and other incentives are included only when supported by an applicable programme or executed agreement. The model distinguishes approved benefits from applications and negotiations, applies the expected timing and recognises performance, reporting or clawback conditions. Incentives can improve economics, but they do not cure an unverified power path, weak tenant demand or an unsupported construction budget.
The development budget is disaggregated into land and site work, utility and substation work, shell construction, electrical distribution, generators, uninterruptible power systems, cooling, controls, security, fire protection, network rooms, fibre extensions, commissioning, professional fees, financing costs, owner-furnished equipment, reserves and contingency. Costs are assigned to the phase that incurs them and reconciled with contractor scopes and utility estimates. Shared campus infrastructure is identified separately so the lender can see how much capital is committed before later-phase demand is proven.
Procurement timing is tested for transformers, switchgear, generators, cooling equipment, control systems and other long-lead components. Vendor quotation validity, manufacturing slots, substitutions, deposits, cancellation terms and storage are relevant to both completion risk and working capital. The schedule links design release, permits, utility work, equipment delivery, construction, integrated systems testing and tenant commissioning. A nominal building completion date does not create revenue if power or commissioned capacity arrives later.
Contract review considers guaranteed maximum price or lump-sum coverage, allowances, owner-supplied items, escalation, change orders, delay allocation, liquidated remedies, performance testing, warranties, bonding and contractor capacity. The feasibility study does not replace counsel or the independent engineer. It reconciles their evidence with the cost and timing assumptions used for underwriting and identifies scope left outside contracted work.
Hyperscale, colocation and enterprise demand are analysed separately because they produce different contracts and risk. A hyperscale facility may offer substantial pre-commitment but concentrate cash flow in one counterparty and a highly specific fit-out. Colocation revenue may be diversified across customers yet require a longer sales ramp, network density and ongoing churn management. Enterprise facilities depend on local and regional requirements, compliance, disaster recovery or owner occupancy and cannot be underwritten from national cloud demand alone.
Vacancy is measured as both physically available and power-ready capacity, with distinctions for facility scale, density, redundancy, network position and lease structure. Announced pipeline is classified by status: proposed, entitled, power-secured, under construction, commissioned or available. This prevents a press release without land, power or financing from being treated as equivalent to near-term competing supply. Preleasing and known tenant commitments are considered where evidence permits, while confidential or unverified market commentary is labelled rather than converted into fact.
Absorption is the pace at which comparable critical load has become contracted and operational in the relevant market and segment. The study examines historical leasing, current requirements, broker and operator evidence, executed subject commitments and the timing of competitive deliveries. Gross leasing is not automatically net absorption, and a backlog of demand may compete for a limited number of sites without flowing to the subject. The stabilisation schedule is therefore tied to phase availability, sales capability and the facility's actual competitive position.
Revenue is built from executed or supportable contract terms: committed critical load, recurring capacity charges, power pass-throughs, cross-connects and other services, commencement dates, escalation, renewal and termination rights. Reimbursed electricity is presented consistently on both sides of the income statement so pass-through revenue does not inflate margin. Operating expenses include utility demand and consumption charges, cooling, staffing, security, maintenance contracts, fuel and testing, network costs, insurance, property taxes, administration and replacement reserves.
Debt-service coverage is measured through construction, phased energisation, lease-up and stabilisation. Interest carry and operating deficits before break-even are included in capital needs. Downside cases combine utility delay, equipment delay, cost overrun, slower absorption, lower contract rates, higher power cost, reduced efficiency, tenant commencement delay, anchor non-renewal and additional capital expenditure. The model also tests whether covenants are met before the final phase, rather than relying on ultimate campus economics to mask weak early coverage.
On a cash-flow-dependent data centre credit, principal underwriting risks include power deliverability, completion and commissioning, tenant concentration, specialised improvements, releasing downtime, technology and density change, energy-price exposure, water and environmental constraints, sponsor liquidity, contractor and supplier performance, incentive clawback and obsolescence. A credible data center feasibility study consultant identifies which risks are transferred by contract, which are mitigated by reserves or guarantees and which remain with the borrower and lender.
Early-stage sponsors comparing locations can use Wert-Berater's site-selection comparison and custom market reports; where the competitive set is principally industrial, the warehouse and industrial market-report practice can provide a separately scoped view of land, pipeline and location evidence. A lender requiring an opinion of collateral value can commission a commercial property appraisal apart from the feasibility assignment. If an eligible owner-user is pursuing SBA financing, the SBA programme feasibility scope applies when the lender or CDC determines a study is needed. For an eligible rural project, the USDA OneRD feasibility service addresses the applicable Agency requirements. These workstreams can use common project documents, but each reaches a distinct conclusion.
Power procurement is often the gating item on a computing campus, and sponsors regularly pair the site with dedicated generation. Where that is contemplated, the generation side is evaluated in our solar farm feasibility study practice.
Legal disclosure. Wert-Berater, Inc. offices are mailing addresses only. Following the COVID-19 pandemic the firm has elected to work remotely; its office locations receive mail and are not staffed for visitors or in-person meetings. Headquarters mailing address: 1968 South Coast Hwy, Ste 2382, Laguna Beach, CA 92651.
Wert-Berater, Inc. is an independent provider of feasibility studies and other related services. The firm does not provide financing or equity investment advice, and does not arrange, broker, or place debt or equity capital of any kind.
All appraisal assignments are performed by Bruce E. Jones, MAI, ASA-GC, BCA, CMEA, a member of the Appraisal Institute since 2006, a staff member of Wert-Berater, Inc. and owner of Special Purpose Realty Valuation.