Our Methodology

Seven Phases. One Methodology. Every Scale.

Data center delivery is sequential: each phase generates decisions that constrain everything after it. The methodology exists to make those decisions with precision — whether the project is a 2MW edge deployment or a 50MW phased campus.

Why Methodology Matters

Most data center projects don't fail during construction. They fail during the decisions that precede construction.

A site selected without power-first analysis locks the project into a 3–5 year utility queue. A design approved without constructability review creates change orders during steel erection. A cooling specification chosen without lifecycle cost analysis costs seven figures over a decade.

The methodology is the countermeasure. It sequences decisions so each phase informs the next, surfaces risks before they become costs, and ensures every stakeholder operates against the same set of constraints. It is not a checklist. It is an operating system for project delivery.

Process as proof

Deliverables, gates, and decision frameworks

When work is under NDA, the credibility test is whether the operating system survives scrutiny from your GC, commissioning agent, and internal engineering team. Below is representative of what we bring into reviews — formats adapt to program scale and contract structure.

Sample deliverable formats

  • Requirements register — Traceable ID per requirement (capacity, redundancy, density, SLA), source (charter, lease, internal ops), verification method, and SD/DD/CD disposition with revision dates.
  • Site evaluation scorecard — Weighted power, fiber, permitting, natural hazard, and labor fields; explicit go/no-go thresholds and documented assumptions for utility verbal vs. written confirmation.
  • Design review memo (per gate) — Issue log with discipline, severity (blocker / major / minor), drawing/spec reference, recommended disposition, and owner decision deadline.
  • Monthly construction report — Schedule variance vs. baseline (critical path highlighted), change-order log with cumulative TIC impact, top three risks with mitigations, and photo-backed QA observations.
  • Commissioning tracker — Script ID, prerequisite systems, pass/fail, deficiency category, responsible party, and re-test date — aligned to IST and turnover gates.

Gate review criteria (examples)

Example criteria used at design and construction gates
GatePass condition
SD releaseSingle-line diagrams and block loads prove N paths for normal and maintenance modes; major equipment spaces and maintenance clearances reserved on plan.
DD releaseEquipment schedules match load growth; selective coordination / arc flash assumptions stated; sequences of operation cover failover and restart; budget estimate within agreed envelope or explicit variance approved.
IFC / GMPSpecs are procurement-ready (alternates bounded); long-lead packages identified with order-by dates tied to schedule; AHJ open items have owner risk register entries.
IST readinessPrerequisites closed (startup reports, controls points proven); labeled failure scenarios approved; owner operators named for witness events.

Decision frameworks we apply

  • Change orders — Classify as true scope addition, spec ambiguity, constructability fix, or expedite; map each to schedule critical path and downstream commissioning before owner approval.
  • Equipment substitution — Score first cost, lead time, efficiency, maintainability, spares commonality, and Cx re-test scope; substitutions that shorten schedule but widen IST risk require explicit owner sign-off.
  • Design freeze vs. utility — If interconnection or capacity letter lags, define which packages may proceed vs. which must wait — documented so the team does not build ahead of a binding utility position.
  • Phasing / tie-in — For live campuses, every energization step lists rollback, who holds the redlines, and communication protocol to operations — no ad-hoc cutovers.
The Framework

Every data center project — regardless of scale, location, or build methodology — moves through the same seven phases. What changes between a 2MW edge facility and a 30MW campus is not the sequence. It is the depth, duration, and complexity within each phase.

1

Discovery & Requirements

2–4 weeks

Requirements document, project charter, success criteria

Before site selection begins, the project needs a single document that defines what success looks like. Capacity requirements. Redundancy targets. Power density. Cooling approach constraints. Timeline. Budget envelope.

This phase prevents the most expensive mistake in data center development: solving the wrong problem precisely.

What we document:

  • Capacity requirements (kW, rack count, density per rack)
  • Redundancy and tier targets
  • Power density and cooling approach constraints
  • Budget envelope and funding timeline
  • Regulatory and compliance requirements
  • Operational model (owner-operated, leased, colocation)
2

Site Selection & Due Diligence

8–12 weeks

Site evaluation matrix, utility analysis, risk assessment

Power is the first filter. Not geography, not cost-per-acre, not proximity to demand. Every site evaluation begins with utility capacity, interconnection timeline, and cost-per-MW of delivered power.

A site with available land but a 4-year utility queue is not a viable site. A site with fast interconnection but insufficient fiber density limits tenant attraction. The evaluation matrix weights these factors against the requirements document from Phase 1.

Evaluation criteria:

  • Utility capacity and interconnection timeline
  • Fiber connectivity (carrier diversity, latency to demand centers)
  • Zoning and permitting pathway
  • Environmental and geotechnical conditions
  • Labor market for construction and operations
  • Natural disaster exposure and insurance implications
  • Tax incentives and jurisdictional considerations

What we hold back for client engagements: Proprietary weighting model and scoring thresholds that determine go/no-go decisions.

3

Design Oversight

4–6 months (SD through CD)

Design review reports at SD, DD, and CD milestones

Design oversight is not design management. The architect and MEP engineer design the facility. We review every deliverable against the requirements document, flag deviations, and ensure the owner's operational intent survives the translation from specification to drawing.

The three most common design failures we catch: undersized electrical infrastructure that limits future density, cooling approaches that optimize first cost at the expense of PUE, and constructability issues that create change orders during build.

Review gates:

  • Schematic Design (SD): Does the concept satisfy the requirements document?
  • Design Development (DD): Are systems sized correctly? Are redundancy paths validated?
  • Construction Documents (CD): Can this actually be built as drawn? Are specifications procurement-ready?

At each gate, we verify:

  • Power distribution architecture and redundancy paths
  • Cooling system capacity and efficiency projections
  • Structural and spatial coordination (will it all fit?)
  • Code compliance and AHJ requirements
  • Constructability and sequencing feasibility
  • Budget alignment (is the design still within the envelope?)
4

Preconstruction & Procurement

2–4 months

GC selection, equipment procurement strategy, baseline schedule

Preconstruction is where the project transitions from paper to commitment. GC selection, equipment procurement, and schedule baselining happen here. The decisions made in this phase determine the financial trajectory of the entire build.

Equipment lead times drive the critical path. A generator with a 40-week lead time ordered 2 weeks late pushes commissioning by 2 weeks — and every week of delayed commissioning is a week of delayed revenue.

What we manage:

  • GC qualification and bid evaluation
  • Equipment procurement strategy (direct-buy vs. GC-furnished)
  • Lead time tracking and procurement sequencing
  • Baseline schedule development with critical path analysis
  • Subcontractor prequalification
  • Insurance and bonding verification
5

Construction Oversight

8–14 months (varies by scale and methodology)

Monthly progress reports, QA documentation, change order log

Construction oversight means presence. Site visits. Submittal reviews. RFI management. Change order evaluation. Safety monitoring. The Owner's Rep is the owner's eyes on the ground — the role that ensures what gets built matches what was designed and what was paid for.

Change orders are the primary financial risk during construction. Not material cost increases. Not labor shortages. Change orders — because they compound. A $50K change order that shifts ductwork creates a $200K cascade through electrical and fire protection. We evaluate every change order against scope, schedule, and downstream impact before the owner approves.

Weekly activities:

  • Site walk and progress documentation
  • Submittal and shop drawing review
  • RFI tracking and resolution
  • Change order evaluation (scope, cost, schedule impact)
  • Schedule update and critical path monitoring
  • Safety and quality assurance observations
  • Monthly cost reporting and budget-to-actual tracking

Cost benchmarks (published ranges):

Construction cost benchmarks by facility scale
ScaleTypical $/MWConstruction Duration
Under 1MW$12–15M/MW6–10 months
1–5MW$10–13M/MW8–14 months
5–20MW$8–11M/MW12–18 months
20–50MW$7–10M/MW14–24 months

These are order-of-magnitude ranges. Actual costs vary by location, tier, cooling approach, and build methodology.

6

Commissioning

6–8 weeks

Cx plan, test scripts, IST results, punch list, turnover documentation

Commissioning is where the data center proves it can perform as designed. It is the most technically demanding phase — and the most commonly under-resourced.

A facility that passes commissioning operates as specified from Day One. A facility that skips or shortcuts commissioning discovers its failures in production — at the cost of uptime, revenue, and reputation.

Commissioning sequence:

  1. Cx Plan Development — Define every system to be tested, every failure scenario to be validated, every performance threshold to be met
  2. Component Testing — Individual equipment startup and verification
  3. Subsystem Testing — Integrated testing within each discipline (electrical, mechanical, controls)
  4. Integrated Systems Testing (IST) — Full-facility testing under simulated load conditions, including failure scenarios
  5. Failure Scenario Validation — Utility loss, generator transfer, cooling failure, switchgear failure — every designed redundancy path is tested
  6. Punch List Resolution — Deficiencies identified, tracked, and resolved
  7. Turnover — Documentation, training, and operational handoff

Commissioning scope by scale:

Commissioning scope and cost by facility scale
ScaleTypical Cx CostCx DurationIST Complexity
Under 5MW$200K–$800K4–6 weeksModerate
5–20MW$800K–$2M6–8 weeksHigh
20–50MW$2M–$4M8–12 weeksVery High
7

Turnover & Day One Operations

2–4 weeks

As-built documentation, O&M manuals, training completion, operational procedures

The project is not complete when commissioning passes. It is complete when the operations team can run the facility independently. Turnover includes as-built documentation, operations and maintenance manuals, vendor warranty consolidation, and operator training.

Turnover deliverables:

  • As-built drawings (architectural, structural, MEP, controls)
  • Operations and maintenance manuals for all major systems
  • Warranty documentation and vendor contact consolidation
  • Spare parts inventory and recommended stock levels
  • Standard operating procedures for normal and emergency operations
  • Training completion records for all operational staff
Scale Adaptation

The seven phases are constant. The methodology within each phase adapts to three variables: project scale, operational model, and build methodology.

Edge Deployments (Under 1MW)

Single-building, often remote. Power procurement is the entire project. Compressed timelines. Modular or prefab builds predominate. Commissioning is streamlined but still rigorous — a 500kW facility with no redundant building has zero margin for error.

Enterprise & AI Infrastructure (1–10MW)

Purpose-built for a single operator. GPU-dense facilities demand liquid cooling analysis. Power density per rack drives the entire design. Timeline pressure is high — AI inference demand doesn't wait for construction delays.

Colocation & Multi-Tenant (5–30MW)

Phased deployment. Tenant-ready shell with fit-out flexibility. The methodology must account for variable density requirements, metering, and common infrastructure shared across tenants. Design oversight focuses on future-proofing.

Campus Developments (20–50MW+)

Multi-building, phased over years. The methodology scales into program management — coordinating multiple design teams, multiple GCs, and phased utility interconnections. A decision in Phase 1 Building A constrains Phase 3 Building D. Context continuity across phases is the primary value of full-lifecycle engagement.

AI in Our Workflow

AI is a practice within the methodology, not a product we sell. It amplifies analysis across every phase — but never replaces the judgment that mission-critical facilities demand.

How we use AI in project delivery:

  • Site evaluation: AI-powered analysis of utility data, interconnection timelines, and regulatory environments to accelerate due diligence
  • Design review: Automated cross-referencing of specifications against requirements documents and industry standards
  • Cost analysis: Pattern matching against historical project data to identify cost anomalies and benchmark deviations
  • Schedule optimization: Critical path analysis with lead time variability modeling
  • Commissioning: Test script generation and IST scenario planning based on system topology

What AI does not do in our workflow: Make decisions. Approve change orders. Accept design compromises. Override engineering judgment. AI surfaces information. People make calls.

We also publish 8 open-source MCP-powered engineering calculation tools — cooling load, power sizing, tier classification, GPU cooling, UPS sizing, and more. These are available to any AI client and any operator. View AI Tools →

Benchmarks

We publish benchmarks because informed owners make better decisions. These are industry ranges derived from project data and published sources — not proprietary client data.

Industry benchmarks for data center project delivery
MetricRangeNotes
Total project cost ($/MW)$7M–$15MVaries by scale, location, tier, cooling
Construction duration6–24 monthsScale and methodology dependent
Commissioning cost$200K–$4MScales with facility complexity
Owner's Rep fee2–4% of project costScope and service tier dependent
Independent OR impact (industry benchmarks)10–15% TICPublished ranges; not client-attributed. Varies by phase entry and governance
Site selection duration8–12 weeksPower availability drives timeline
Design phase (SD through CD)4–6 monthsComplexity dependent
Commissioning duration4–12 weeksScale and IST complexity dependent
Next step

Discuss your project.

One conversation. No obligation. We'll tell you where in the seven phases your project sits — and what the next decision should be.