Understand Data Center feasibility: demand, site, power, connectivity, deployment models, CAPEX, OPEX, schedule, risks, evidence levels and Go/Hold/No-Go gates.
Check it out!
A Data Center feasibility study verifies whether demand, site, power, connectivity, deployment model, budget, schedule, and risks form an executable alternative before design development and construction contracting. Its objective is not to confirm a decision already made, but to produce sufficient evidence to recommend proceeding, revising, holding, or abandoning the alternative.
The analysis should begin with the business problem and the ICT load that must be served. It then converts those needs into comparable scenarios for capacity, architecture, location, and operations. Power and connectivity are usually the most critical constraints, but land, permitting, water, cooling, logistics, security, supply chain, schedule, and expansion capacity can make an apparently attractive development unfeasible.
The appropriate outcome is not a promise that the project will be approved or implemented without changes. It is a documented basis for deciding Go, Hold, or No-Go, stating conditions, and defining which specialized studies need to be commissioned at the next gate.
Technical summary
| Question | Answer |
| What does the study verify? | Whether demand, site, power, telecommunications, architecture, costs, schedule, and risks are compatible |
| When should it be performed? | Before final property acquisition, detailed design, and major contracting commitments |
| What is the primary output? | A Go, Hold, or No-Go recommendation accompanied by assumptions, risks, conditions, and a plan for further studies |
| Does feasibility replace design? | No. It establishes the basis for conceptual, basic, and detailed design |
| Is an informal utility inquiry sufficient? | No. It is necessary to distinguish a preliminary indication, connection study, quotation, commitment, and formal schedule |
| Do two carriers guarantee diversity? | No. Routes may share ducts, poles, backbones, or entry points |
| Is the lowest CAPEX always the best alternative? | No. Schedule, OPEX, risk, expansion, responsibilities, and the cost of downtime also influence the decision |
What is a Data Center feasibility study?
It is a multidisciplinary analysis performed before detailed design to verify whether an investment alternative can meet owner and user requirements within acceptable limits of cost, schedule, risk, and operations. It can be applied to a new campus, corporate Data Center, colocation, Edge, Micro Data Center, modular deployment, expansion, or modernization of an existing facility.
Feasibility sits between the intent to invest and definition engineering. It structures the problem, consolidates data, eliminates incompatible options, compares scenarios, and reduces uncertainty. The earlier a critical constraint is identified, the lower the cost of abandoning or reformulating the alternative tends to be.
Feasibility is not a commercial justification
The study loses value when it is conducted only to validate a selected site, manufacturer, architecture, or previously defended budget. The team should be free to demonstrate that the site lacks power within the required schedule, that connectivity is not diverse, that phasing creates idle infrastructure, or that colocation presents lower exposure for a given demand.
The conclusion needs to reflect the available data. When information has not been confirmed, the report should record the assumption, its impact, and the action required to close it.
Difference between feasibility study, due diligence, and design
| Stage | Main question | Expected depth | Output |
| Screening | Which alternatives deserve further assessment? | Public data, assumptions, and initial inquiries | Shortlist of sites or models |
| Due diligence | What constraints and liabilities exist? | Documents, surveys, visits, and validations | Diagnosis, risks, and open items |
| Feasibility study | Can the alternative meet the investment requirements? | Integrates requirements, site, architecture, costs, schedule, and risks | Go, Hold, or No-Go |
| Conceptual design | Which architecture should be developed? | Systems, capacities, blocks, and interfaces | Basis of Design and reference solution |
| Basic and detailed design | How should it be contracted and built? | Calculations, specifications, drawings, and details | Implementation documentation |
The stages may overlap, but they should not be confused. Land due diligence alone does not determine whether the development is economically feasible. A preliminary layout does not prove electrical connection. A parametric estimate does not replace risk and schedule analysis.
When should the study be performed?
The best time is before making commitments that are difficult to reverse. This includes final land purchase, long-term lease agreements, reservation of critical equipment, permit applications based on an architecture that is still unstable, or contracting detailed design.
In larger developments, feasibility can be organized by gates:
- Opportunity gate: confirms the business problem, demand, and strategic alternatives.
- Location gate: selects regions and sites with minimum conditions for power, telecommunications, permitting, and expansion.
- Architecture gate: compares owned, colocation, modular, Edge, cloud, or hybrid models.
- Investment gate: consolidates CAPEX, OPEX, schedule, risks, and sensitivity.
- Development gate: authorizes conceptual design, specialized studies, and contracting of the next phase.
The decision may be revisited when utilities, public authorities, operators, or field surveys update critical information.
Define the problem before evaluating the site
Many studies begin with the available property. This sequence can lead to a solution built around site constraints without checking whether another model would better serve the business. The starting point should be demand.
Project objective
It is necessary to clarify whether the infrastructure will be used for corporate services, continuity, industrial processing, colocation, cloud, AI, telecommunications, regional low latency, data sovereignty, or real-estate development. Different objectives change density, connectivity, availability, security, expansion, and expected return.
ICT load profile
Demand should be represented by power, racks, densities, storage, network, growth, and criticality. Floor area is a consequence, not a primary requirement.
| Parameter | Feasibility question |
| Initial ICT load | How much needs to go into operation in the first stage? |
| Ultimate capacity | What is the realistic growth horizon? |
| Average and maximum density | Will there be traditional, HPC, or AI zones? |
| Occupancy rate | How long will it take for each block to be utilized? |
| Time profile | Is the load stable, seasonal, or step-growing? |
| Availability | What interruption impact is acceptable for each service? |
| Data and network | What volumes, latencies, routes, and integrations are required? |
| Operations | Who will maintain the facility and with which resources? |
When demand is still uncertain, the study should work with low, base, and high scenarios. Designing directly for the maximum scenario may bring CAPEX forward, increase losses, and create idle assets. Designing only for the initial scenario may block expansion.
Availability requirements
Availability should be defined for services, not only for equipment. Application architecture, ICT redundancy, physical infrastructure, operations, and continuity across sites need to be related. The article on Tier I, II, III, and IV in Data Centers explains why redundancy and classification are not synonymous.
Feasibility study work structure
A consistent methodology organizes disciplines into parallel workstreams with shared assumptions and interfaces.
| Workstream | Core questions | Primary evidence |
| Business and demand | Why invest, how much, and when? | Capacity plan, requirements, and scenarios |
| Location | Can the site support implementation and expansion? | Documents, maps, surveys, and visits |
| Power | Is there sufficient power, schedule, and connection architecture? | Letters, studies, diagrams, and formal inquiries |
| Telecommunications | Is there capacity, latency, and diversity? | Routes, carriers, PoPs, and infrastructure evidence |
| Infrastructure | Which architectures meet the requirements? | Diagrams, blocks, layouts, and performance criteria |
| Permitting | Which approvals and constraints condition the schedule? | Certificates, inquiries, and permit matrix |
| Economics | How much does it cost to build, operate, and expand? | CAPEX, OPEX, TCO, and sensitivities |
| Risks | What can prevent or degrade the investment? | Register, owners, responses, and conditions |
| Implementation | What is the critical path? | Schedule, lead times, and contracting strategy |
The workstreams should not operate as independent reports. A density change affects power, cooling, water, structure, equipment, and CAPEX. A site change affects latency, permitting, logistics, and connection schedule.
Evidence and confidence levels
The report should clearly distinguish confirmed information, estimates, and assumptions. A useful practice is to classify evidence:
| Level | Example | Permitted use |
| E0 — assumption | Verbal information or internal assumption | For screening and scenarios only |
| E1 — indication | Public map, catalog, or non-binding statement | Preliminary comparison |
| E2 — documentary evidence | Document from the owner, carrier, or authority | Basis for the study, with caveats |
| E3 — technical validation | Survey, test, study, or formal inquiry | Conditional decision or further detailing |
| E4 — commitment | Contract, permit, opinion, or applicable formal condition | Basis for investment commitment |
A conclusion may be positive even with open items, provided they are explicit and associated with gates. The mistake is to present an E0 assumption as if it were an E4 commitment.
Power assessment for Data Centers
Power often determines location, capacity, phasing, and schedule. The study needs to assess the external system and the internal infrastructure required to convert contractible power into usable ICT load.
Electrical load model
Total electrical demand is not equal to ICT load. Cooling, losses, lighting, security, pumping, utilities, and margins must be included. At a preliminary stage, parametric factors may be used, but the report should state their basis.
A simple model relates:
Estimated total power = ICT load × infrastructure factor + auxiliary loads not represented in the factor.
This calculation does not replace design. It serves to size orders of magnitude, compare alternatives, and request coherent information from the utility.
What to verify with the utility or connection agent
The analysis should go beyond the question “is power available?”. It is necessary to understand the connection point and voltage, power available by phase, works and reinforcements, responsibilities for substations and lines, easements, study and energization lead times, reliability, power quality, tariff structure, and expansion potential.
A preliminary indication should not be confused with a service guarantee. In critical projects, the land decision and schedule need to state which document supports the electrical assumption.
Firm, rated, and usable capacity
| Concept | Meaning |
| Rated capacity | Nameplate or design limit of the assets |
| Available capacity | Uncommitted portion under the condition analyzed |
| Contractible capacity | Power that can be formally supplied |
| ICT capacity | Portion allocated to technology equipment |
| Usable capacity | Possible load after margins, redundancy, and operating limitations |
| Future capacity | Expansion dependent on works, permits, or reinforcements |
Confusing these values can overestimate the number of racks or marketable MW.
Preliminary power-supply architecture
Feasibility should compare connection and distribution alternatives at a level sufficient to estimate area, cost, schedule, and risk. This may involve one or more feeds, substations, emergency generation, UPS, batteries, A/B distribution, and capacity blocks.
It is not necessary to finalize the electrical design, but it is necessary to identify the elements that dominate the investment and critical path. The future article Data Center electrical architecture: N, N+1, 2N, and A/B distribution will explore these topologies in more depth.
Generation, fuel, and autonomy
Generator sets require space, fuel routes, storage, ventilation, exhaust, noise control, emissions management, permitting, and testing. Autonomy should reflect the risk of utility interruption and refueling time. Urban, industrial, and remote sites present different constraints.
The presence of generators does not eliminate the need to analyze fuel logistics during regional events, blocked access, or common-supplier failure.
Power quality
Voltage sags, interruptions, harmonics, imbalances, and grid events can affect UPS systems, protection, generators, and equipment. When power quality is a relevant constraint, the study should recommend a measurement campaign or specific analysis rather than assume adequate behavior.
Power needs to be treated as a formal feasibility condition, not as a preliminary indication.
A3A Engenharia structures demand, connection scenarios, reinforcements, phasing, and the evidence required for the investment decision.
Telecommunications connectivity and diversity
Connectivity should be assessed as both physical infrastructure and an operational service. The number of carriers is only an initial indicator.
Network demand
The study needs to relate applications, data volume, east-west and north-south traffic, replication, backup, internet access, cloud, interconnection, and growth. For Edge and AI, latency, jitter, and throughput may be decisive.
Routes and points of presence
The investigation should identify where fibers come from, where they enter the property, which ducts and poles they use, which backbones feed the carriers, and where relevant PoPs are located. Two contracts may share the same trench or regional infrastructure.
| Criterion | Due diligence question |
| Carriers | Are there providers with compatible capacity and support? |
| Routes | Are the paths physically diverse to the required boundary? |
| Entries | Does the site allow separate and protected entries? |
| Backbone | Is there regional capacity and planned expansion? |
| Latency | Can critical destinations be served? |
| Interconnection | Are there relevant IXPs, cloud on-ramps, MMRs, or ecosystems? |
| Rights of way | Do ducts, poles, and corridors depend on third parties or permits? |
| Schedule | Can route construction fit within the schedule? |
Carrier-neutrality and ecosystem dependency
For colocation, the presence of multiple carriers and interconnections may be central to the value proposition. For a corporate Data Center, two controlled routes may be sufficient. Feasibility should assess the actual requirement, avoiding copying a market architecture that is unrelated to the operating model.
The article Colocation Data Center: how it works and how to evaluate a provider presents additional criteria for contracted services.
Continuity during WAN loss
In Edge and Micro Data Centers, the infrastructure may exist precisely to continue operating without a central connection. The study should define which functions remain local, how data are stored, how synchronization occurs, and which conflicts may arise when connectivity returns.
Site and physical implementation
A Data Center site needs to support the facility throughout its lifecycle, not only the first building. A large gross area may conceal implementation restrictions, easements, drainage constraints, setbacks, topography, or non-buildable zones.
Land documentation and control
Title records, ownership, encumbrances, boundaries, access, existing easements, and rights required for power, telecommunications, water, and drainage should be verified. Legal matters require appropriate specialists; engineering identifies interfaces and impacts on the development.
Land use and permitting
The study should map zoning, building parameters, environmental restrictions, heritage constraints, aeronautical requirements where applicable, fuel storage, noise, emissions, vegetation removal, water resources, and fire-authority requirements.
The permit matrix should indicate the authority, document, dependencies, estimated lead time, required data, and risk of conditions.
Topography, geotechnics, and drainage
Slope influences earthworks, access, retaining structures, and utilities. Geotechnical conditions may change foundations and schedule. Inadequate drainage creates a direct risk to critical infrastructure.
During screening, existing maps and information can eliminate clearly unsuitable sites. Before final investment, topographic surveys, geotechnical investigations, and hydrological studies need to be commissioned according to risk.
Flooding and natural hazards
The analysis should consider watercourses, elevations, regional drainage, event history, mass movements, winds, lightning, wildfire, extreme temperatures, and other threats relevant to the site. Generic classifications do not replace a specific study when the risk is material.
Logistics and accessibility
Transformers, generators, chillers, modules, and large equipment require roads, turning radii, bridges, gates, unloading areas, and lifting equipment. The study should also assess staff access during emergencies and weather events.
Security and surroundings
Industrial activities, railways, highways, airports, hazardous-material storage, conflict areas, vandalism, and neighboring occupancies may influence risk. Physical security, lighting, perimeter, and setbacks need to be coordinated with the site.
Expansion area
The preliminary masterplan should reserve space for buildings, substations, generation, tanks, utilities, telecommunications, roads, drainage, workshops, storage, and construction areas for future phases.
An apparently free area may become unusable once infrastructure corridors, setbacks, easement strips, and safety distances are applied.
Climate, water, and thermal solution
Thermal feasibility depends on load, density, climate, and resource availability. Outdoor temperature and humidity influence efficiency and system selection. Altitude affects equipment performance. Dust, salinity, and corrosive gases affect materials and filtration.
Water availability and quality
Evaporative systems, towers, and some liquid-cooling architectures may depend on water. The study should assess source, flow rate, quality, seasonality, permits, treatment, discharge, and competition with other users.
An energy-efficient solution may increase water consumption or operational complexity. The comparison needs to use equivalent indicators and boundaries.
Heat rejection
Sizing internal units is not enough. The site needs to reject heat under design conditions and during expansion phases. Space for chillers, dry coolers, towers, CDUs, piping, and maintenance should be reserved from feasibility onward.
High density and AI
AI loads may require higher densities, high-speed networks, and liquid cooling. The study should verify whether this demand is confirmed, likely, or merely possible. Designing the entire development for maximum density may increase CAPEX and reduce efficiency.
One alternative is to define zones, blocks, or phases with different envelopes while preserving interfaces for technology evolution.
Comparison of deployment models
The feasibility study should not assume that building is the only answer. Different models transfer responsibilities and risks.
| Model | Primary advantage | Primary risk or limitation |
| Owned Data Center | Control of architecture, assets, and operations | CAPEX, schedule, staffing, and obsolescence |
| Colocation | Contracted infrastructure and operations | Recurring cost, dependency, connectivity, and exit |
| Cloud | Elasticity and managed services | Variable costs, dependency, data, and latency |
| Hybrid | Distribution according to workload profile | Integration, governance, and complexity |
| Edge | Low latency and local autonomy | Distributed operations and standardization |
| Micro Data Center | Compact deployment near the load | Expansion, maintenance, and local environment |
| Modular Data Center | Phasing and prefabrication | Interfaces, logistics, and technology lock-in |
| Build-to-suit | Capacity developed for specific requirements | Contract, schedule, and operator dependency |
The comparison should use the same demand, horizon, availability, and cost boundary. Comparing elastic cloud with an owned Data Center sized for ultimate capacity without considering utilization produces a distorted result.
The articles on Modular Data Center, Edge Data Center and Micro Data Center explore these alternatives in more depth.
Preliminary architectures and scenarios
After eliminating incompatible alternatives, the team develops conceptual scenarios. The objective is to estimate capacity, area, systems, cost, schedule, and risks without prematurely advancing detailed design.
Capacity blocks
Deployment can be organized into ICT-load blocks associated with power, cooling, and space. Block size should balance economies of scale, delivery lead time, utilization, and flexibility.
| Decision | Consequence of a large block | Consequence of a small block |
| Initial CAPEX | Greater upfront investment | Better alignment with demand |
| Procurement efficiency | Potential economies of scale | More contracts and interfaces |
| Flexibility | Less adaptability | More phases |
| Commissioning | Larger campaign | Repeated testing |
| Operations | Fewer variants | More transitions between phases |
Shared infrastructure
Substation, generation, utilities, security, and telecommunications may be brought forward for multiple phases. The study should compare the cost of anticipating assets with the risk of expanding later in an active campus.
Expansion triggers
Each phase needs to begin before saturation. Triggers may combine contracted capacity, occupancy, commercial demand, equipment lead time, connection lead time, and operating margin.
Constructability in an active campus
Future phases need to be built without blocking access, interrupting fibers, reducing redundancy, or exposing operating rooms to dust, water, and construction risks. The preliminary masterplan should separate construction and operational flows.
The approved alternative needs to be converted into technically controlled requirements, architecture, and interfaces.
Capacity blocks, masterplan, power, cooling, telecommunications, security, automation, and acceptance criteria should preserve the assumptions validated during feasibility.
CAPEX, OPEX, and lifecycle cost
Economic modeling during feasibility works with ranges and assumptions consistent with the level of maturity. It should not claim detailed-estimate accuracy.
Preliminary CAPEX
It may include site acquisition or adaptation, earthworks, civil works, external power, substation, generation, UPS, batteries, cooling, telecommunications, security, fire protection, automation, racks, design, management, testing, permits, contingencies, and escalation.
OPEX
It should consider energy, demand charges, water, maintenance, parts, fuel, telecommunications, licenses, security, insurance, staffing, contracts, battery and equipment replacement, in addition to expenses specific to the contracted model.
TCO and horizon
Total cost should be compared over a horizon consistent with asset lives and the contract. Colocation may have lower CAPEX and higher recurring costs. Owned infrastructure may require high initial investment and reinvestment before the end of the horizon.
Cost of downtime and delay
Lost revenue, penalties, production interruption, delayed launch, or computing unavailability may exceed differences in CAPEX. When relevant, these impacts should appear in risk or sensitivity analysis without creating false precision.
Sensitivity
Critical variables can be tested in scenarios: electrical-connection lead time and cost, load growth, commercial occupancy, energy price, exchange rate, ICT density, permit delays, cost of capital, water availability, and technology change.
The analysis shows which assumptions dominate the outcome and where investing in additional information is worthwhile.
Schedule and critical path
The feasibility schedule should integrate external and internal activities. In many projects, electrical connection, permitting, and long-lead equipment determine the operational date.
High-level structure
- Requirements and demand scenarios.
- Regional screening and site shortlist.
- Power and telecommunications inquiries.
- Technical, land, and regulatory due diligence.
- Preliminary architectures and masterplan.
- CAPEX, OPEX, and phasing strategy.
- Risk register and sensitivities.
- Recommendation and decision gate.
- Specialized studies and conceptual design.
- Contracting, construction, commissioning, and acceptance.
Long-lead items
Transformers, generators, UPS systems, switchboards, chillers, liquid-cooling equipment, and automation components may require early procurement. The strategy needs to avoid purchasing before requirements stabilize while also recognizing when waiting for detailed design would jeopardize the schedule.
Permits and dependencies
The schedule should show precedences: an environmental study may depend on the masterplan; the connection may require a defined area; a fuel permit depends on volumes and layout; equipment procurement depends on interfaces and test criteria.
Risk register and analysis
Risk should not be a generic list at the end of the report. Each event needs a cause, consequence, probability, impact, owner, response, due date, and closure evidence.
| Category | Risk example | Possible response |
| Demand | Occupancy below the base scenario | Phase investment and review triggers |
| Power | Grid reinforcement delays energization | Alternative site, temporary generation, or reduced phase |
| Telecom | Routes share infrastructure | Build an alternative entry or accept documented risk |
| Site | Geotechnical investigation identifies low-capacity soil | Redesign foundations, revise CAPEX, or change site |
| Environmental | Water restriction limits technology | Select a dry or hybrid architecture |
| Permitting | Permit conditions change the layout | Reserve areas and bring inquiries forward |
| Supply chain | Transformer cannot meet the schedule | Qualify alternatives and reserve manufacturing capacity |
| Technology | AI density changes during development | Controlled variants and flexible zones |
| Operations | There is no qualified team in the region | Remote model, training, and specialized contracts |
| Economic | Exchange-rate changes alter CAPEX | Contingency, localization, and phased purchasing |
Probability and impact
Qualitative scales may be sufficient during screening. For larger decisions, the study may estimate impact on cost, schedule, capacity, safety, and reputation. Extreme risks should not be diluted by a simple average.
Correlated risks
Delayed power may postpone construction, occupancy, and revenue. Water shortages may change cooling, CAPEX, and permitting. The matrix should record dependencies and combined scenarios.
Gate conditions
A risk may block approval until specific evidence is obtained, such as a formal commitment on power and schedule, proof of a fiber route, geotechnical investigation, environmental consultation, land-use authorization, validation of logistics access, or approval of the financial model.
Risks, conditions, and decisions need to remain traceable throughout development and implementation.
Owner’s Engineering preserves requirements, reviews designs, controls interfaces, tracks evidence, and supports technical gates through acceptance.
Site screening and comparison
When there are several locations, the methodology needs to avoid scoring that hides disqualifying criteria. Minimum criteria are applied first; approved sites are then compared using weights.
Disqualifying criteria
They may include inability to connect within the required schedule, legal restrictions, non-mitigable flood risk, lack of access, insufficient area, or inability to establish telecommunications routes.
Weighted criteria
| Criterion | Illustrative weight | Observation |
| Power and connection schedule | 25% | Should use comparable evidence |
| Telecommunications | 15% | Capacity, latency, and diversity |
| Site and expansion | 15% | Usable area, topography, and phases |
| Permitting and environment | 15% | Schedule, water, noise, and conditions |
| Natural and human-induced risks | 10% | Exposure and mitigation |
| Logistics and supply chain | 10% | Access, labor, and equipment |
| CAPEX and OPEX | 10% | Same analysis boundary |
The weights are only illustrative. A regional Edge facility, a hyperscale campus, and a corporate Data Center have different priorities.
Do not turn uncertainty into a high score
When a site lacks information, the cell should not receive an average score for convenience. Uncertainty needs to be recorded and, if material, treated as a risk or disqualifying open item.
Simplified comparison example
Consider three alternatives for an initial 1 MW ICT load, expandable to 4 MW. The data below are illustrative.
| Criterion | Site A | Site B | Site C |
| Initial power | Available with local works | Depends on regional reinforcement | Available |
| Schedule for 4 MW | 30 months | 48 months | 36 months |
| Fiber | Two verified routes | Two carriers, same route | One current route and a second planned |
| Site | Large, moderate drainage constraints | Excellent topography | Limited expansion area |
| Water | Available with seasonal restriction | Restricted | Available |
| Permitting | Compatible, studies pending | Established industrial zone | Potential noise conflict |
| Initial CAPEX | Medium | Low | High |
| Dominant risk | Water and drainage | Power and telecom | Expansion and surroundings |
The analysis should not automatically choose the lowest CAPEX. Site B may lose attractiveness if the connection schedule makes operations unfeasible. Site C may serve the first phase but block growth. Site A may remain preferred subject to a water and drainage solution.
Feasibility study deliverables
The deliverable set should be proportionate to the stage but must allow the decision to be audited.
| Deliverable | Expected content |
| Requirements report | Objective, demand, scenarios, and performance criteria |
| Assumption register | Source, evidence level, owner, and validity |
| Site diagnosis | Site, power, telecom, environment, logistics, and risks |
| Comparison matrix | Disqualifiers, weights, scores, uncertainties, and justifications |
| Conceptual architectures | Blocks, systems, phases, interfaces, and capacity |
| Preliminary masterplan | Implementation, expansion, access, and utilities |
| CAPEX and OPEX | Ranges, bases, contingencies, and horizon |
| Schedule | Critical path, permits, connection, and long-lead items |
| Risk register | Causes, impacts, responses, owners, and gates |
| Recommendation | Go, Hold, or No-Go and plan for the next phase |
Go
The alternative has sufficient evidence and compatible risks to advance, normally with conditions and further studies.
Hold
The decision should wait for critical information, scope reformulation, third-party commitment, or additional comparison. Hold does not mean abandonment; it means advancing now would expose the investment to excessive uncertainty.
No-Go
The alternative does not meet an essential requirement or presents incompatible cost, schedule, or risk. The report should explain whether another location, scale, or model can be evaluated.
Study limitations
Feasibility does not replace a formal electrical-connection study, topographic survey, geotechnical investigation, hydrological and environmental studies, legal opinion, conceptual, basic, or detailed design, detailed calculations, route confirmation, permits, manufacturer validations, FAT, SAT, or commissioning.
The report should indicate the expected accuracy of each estimate and the disciplines responsible for further development.
Common mistakes
| Mistake | Consequence |
| Selecting the site before confirming power | Land asset without compatible capacity or schedule |
| Using rated power as ICT capacity | Overestimation of the development |
| Accepting two carriers as diversity | Hidden external single point of failure |
| Defining capacity by area | Architecture disconnected from load and density |
| Comparing scenarios with different SLAs | Invalid economic decision |
| Treating a parametric estimate as a fixed price | Underestimation of contingencies and interfaces |
| Ignoring expansion during the masterplan | Premature retrofit or blocked growth |
| Assessing PUE without water and utilization | Incomplete view of efficiency |
| Leaving permitting until detailed design | Delay and redesign |
| Recording risks without owners and gates | Matrix with no effect on the decision |
| Purchasing equipment before requirements | Technology lock-in and costly changes |
| Concluding Go without declaring open items | Commitment based on assumptions |
Feasibility checklist
- Which business problem justifies the investment?
- Which loads, users, and data will be served?
- What are the initial, peak, and ultimate ICT power requirements?
- Which densities and technologies are likely?
- What availability is required for each service?
- Which deployment models should be compared?
- What evidence supports demand and the occupancy rate?
- Is contractible electrical capacity available within the required schedule?
- Which external works, easements, and permits does the connection require?
- Are the telecommunications routes physically diverse?
- Does the site have sufficient usable area and expansion capacity?
- What flood, soil, climate, and surrounding-area risks exist?
- Can water availability and heat rejection support the architecture?
- Which permits condition the schedule?
- How will capacity blocks and expansion triggers be divided?
- Which equipment and activities dominate the critical path?
- Do CAPEX, OPEX, and TCO use equivalent boundaries?
- Which assumptions have weak evidence?
- Which risks block the gate until closed?
- Does the recommendation clearly define Go, Hold, or No-Go and the next phase?
A3A Engenharia Consulting Engineering scope
A3A Engenharia develops feasibility studies for new Data Centers, expansions, modernizations, campuses, corporate environments, Edge, colocation, and modular solutions. The work integrates business requirements, critical infrastructure, site assessment, and decision governance.
The scope may include requirements gathering, regional and property screening, technical due diligence, preliminary power and telecommunications inquiries, capacity analysis, masterplan, architecture scenarios, CAPEX, OPEX, schedule, risk register, and definition of the next-phase scope.
The commercial service is presented in Data Center Feasibility Study. After the decision, the approved alternative can advance to Data Center Design and implementation governance through Owner’s Engineering for Data Centers.
Technical summary
A Data Center feasibility study turns an investment intention into a traceable decision. It integrates demand, site, power, telecommunications, cooling, water, permitting, architecture, capacity, costs, schedule, and risks, distinguishing confirmed evidence from assumptions.
Power and connectivity require documentary validation and analysis of physical paths. The site needs to be assessed for usable area, expansion, risks, access, easements, and permits. Owned, colocation, cloud, Edge, Micro Data Center, and modular scenarios should be compared using equivalent requirements and boundaries.
The recommendation may be Go, Hold, or No-Go. A technically responsible outcome does not promise the absence of risks: it defines which risks are acceptable, which need mitigation, and which evidence must be obtained before the next gate.
Technical references
[1] INTERNATIONAL ORGANIZATION FOR STANDARDIZATION. ISO/IEC 22237-1:2021 — Information technology — Data centre facilities and infrastructures — Part 1: General concepts. Geneva: ISO, 2021.
[2] INTERNATIONAL ORGANIZATION FOR STANDARDIZATION. ISO/IEC 22237-2:2024 — Information technology — Data centre facilities and infrastructures — Part 2: Building construction. Geneva: ISO, 2024.
[3] INTERNATIONAL ORGANIZATION FOR STANDARDIZATION. ISO/IEC 22237-3:2021 — Information technology — Data centre facilities and infrastructures — Part 3: Power distribution. Geneva: ISO, 2021.
[4] INTERNATIONAL ORGANIZATION FOR STANDARDIZATION. ISO/IEC 22237-4:2021 — Information technology — Data centre facilities and infrastructures — Part 4: Environmental control. Geneva: ISO, 2021.
[5] TELECOMMUNICATIONS INDUSTRY ASSOCIATION. ANSI/TIA-942-C — Telecommunications Infrastructure Standard for Data Centers. Arlington: TIA, 2024.
[6] BICSI. ANSI/BICSI 002-2024 — Data Center Design and Implementation Best Practices. Tampa: BICSI, 2024.
[7] UPTIME INSTITUTE. Tier Standard: Topology for Data Center Site Infrastructure. Uptime Institute.
[8] UPTIME INSTITUTE. Tier Standard: Operational Sustainability. Uptime Institute.
[9] ASHRAE. Thermal Guidelines for Data Processing Environments. Atlanta: ASHRAE.
[10] INTERNATIONAL ENERGY AGENCY. Energy and AI. Paris: IEA, 2025.
[11] INTERNATIONAL ORGANIZATION FOR STANDARDIZATION. ISO 31000:2018 — Risk management — Guidelines. Geneva: ISO, 2018.
[12] A3A ENGENHARIA. Data Center Feasibility Study: technical, economic, and risk analysis. Ponta Grossa: A3A Engenharia.
Frequently asked questions
It is a multidisciplinary analysis that verifies whether demand, site, power, connectivity, architecture, costs, schedule, and risks form an executable alternative before detailed design and implementation.
Before final land acquisition, detailed design, and major contracting commitments. It can be updated as critical information is confirmed.
No. Feasibility selects and conditions the alternative. Conceptual, basic, and detailed design develop calculations, systems, specifications, drawings, and implementation criteria.
It is necessary to model ICT and infrastructure load, formally consult the utility or connection agent, and assess connection point, voltage, reinforcements, schedule, costs, reliability, and expansion potential.
No. Carriers may share ducts, poles, backbones, or entry points. Diversity needs to be physically verified to the required boundary.
Usable area, expansion, documentation, zoning, topography, geotechnics, drainage, flooding, access, easements, surroundings, security, permitting, water, and logistics.
The scenarios need to use the same demand, availability, horizon, and cost boundary, including CAPEX, OPEX, risks, responsibilities, expansion, connectivity, and exit strategy.
Go authorizes advancement subject to conditions; Hold suspends the decision until critical information or risks are closed; No-Go rejects the analyzed alternative or recommends changing site, scale, or model.
The estimate is preliminary and should present ranges, bases, contingencies, and level of maturity. It does not have the accuracy of an estimate based on detailed design and firm proposals.
Requirements, assumption register, site diagnosis, comparison matrix, conceptual architectures, preliminary masterplan, CAPEX, OPEX, schedule, risk register, and recommendation for the next gate.
Additional technical resources
Data Center fundamentals and models
- Data Center: what it is, how it works, and which systems make up the infrastructure
- Tier I, II, III, and IV in Data Centers
- Data Center market
Deployment alternatives
- Modular Data Center: types, design, and risks
- Edge Data Center: applications and architecture
- Micro Data Center: applications and specification
- How to evaluate a Colocation provider
- Hyperscale Data Center: what it is and how it works
Feasibility, design, and governance
- Data Center Feasibility Study
- Data Center Design
- Owner’s Engineering for Data Centers
- Integrated Engineering for Data Centers
Power, cooling, and capacity
Telecommunications, security, and fire protection
- Networks and Telecommunications for Data Centers
- Physical Security for Data Centers
- Fire Detection and Suppression in Data Centers
Implementation, testing, and modernization
- Data Center Commissioning and Acceptance
- Diagnosis and Modernization of Data Centers and Computer Rooms
