Senior Data Center Capacity Delivery Manager

Senior Data Center Capacity Delivery Manager

When to Use

• Turn an approved capacity initiative into a dated delivery plan (kW/MW, racks, go-live)
• Run critical path across power, cooling, fit-out, network demarc, and acceptance tests
• Coordinate vendors: GC, MEP, colo, utility, carriers, commissioning agents
• Define readiness gates: construction complete → commissioned → rack-ready → workload-ready
• Track capacity at risk — slip vs demand date from engineering or portfolio
• Manage RAID and weekly status for capacity delivery programs
• Hand off to facilities ops and cluster-deployment-engineer when capacity is install-ready
• Support multiple concurrent hall or row expansions with dependency map

When NOT to Use

• Multi-site capex prioritization and funding bands → data-center-portfolio-planning-execution-lead
• Detailed hall design, single-line diagrams, tier selection → data-center-design-execution-lead
• Host consolidation, GPU forecast, stranded kW optimization → data-center-compute-supply-efficiency
• Cross-team software integration program → technical-program-manager
• Terraform, cloud burst architecture → infrastructure-engineer

Related skills

Core Workflows

1. Delivery charter

Scope: site, increment (e.g. 2 MW, 80 racks), demand date, budget envelope, DRIs.

See `references/delivery_charter.md`.

2. Capacity delivery plan

Milestones from design release through rack-ready and optional workload-ready.

See `references/capacity_delivery_plan.md`.

3. Workstream coordination

Power, cooling, construction, network, security, DCIM — integrated schedule.

See `references/workstream_coordination.md`.

4. Readiness gates

Objective criteria before declaring capacity available for install or production.

See `references/readiness_gates.md`.

5. Vendor and construction management

RFI/submittal cadence, site safety, change orders, acceptance.

See `references/vendor_construction.md`.

6. Reporting and escalation

Weekly RAG, MW/rack burn-up, top blockers, decisions for steering.

See `references/reporting_escalation.md`.

Output standards

• Integrated master schedule with critical path identified
• Capacity burn-up: planned vs actual kW/racks by week
• RAID with owner and target date; escalate reds within 48h
• Readiness checklist signed by facilities and IT infrastructure
• Assumptions on utility lead times and colo lead times explicit

When to load references

• Charter → references/delivery_charter.md
• Schedule and milestones → references/capacity_delivery_plan.md
• Workstreams → references/workstream_coordination.md
• Gates → references/readiness_gates.md
• Site and vendors → references/vendor_construction.md
• Status pack → references/reporting_escalation.md

Capacity Delivery Plan

Milestone chain (typical)

Not every increment needs all steps (e.g. colo turn-key); tailor.

Critical path

Identify longest pole:

• Utility feed and switchgear
• Cooling plant or colo build SLA
• Carrier fiber lead time
• Commissioning window and authority having jurisdiction

Update critical path when change orders land.

Capacity burn-up

Track weekly:

At risk when actual lags planned with no recovery plan in 2 weeks.

Schedule integration

Single master schedule (not separate Gantt per vendor without links):

• Design releases
• Procurement
• Construction
• Network
• Commissioning
• IT install (dependency only)

Demand date negotiation

If slip inevitable:

1. Quantify impact on product/engineering (GPU queue, colo contract) 2. Options: partial MW, phased rows, cloud burst bridge 3. Recommend to data-center-portfolio-planning-execution-lead

Delivery Charter

Charter fields

Scope boundaries

Success criteria

• Rack-ready on or before demand date (or approved re-baseline)
• Acceptance tests passed; as-builts in DCIM
• Operations runbook handed to facilities
• No open red safety or compliance items

Interfaces

Change control

Capacity, date, or budget change → steering or portfolio lead approval; update master schedule within 2 business days.

Readiness Gates

Gate definitions

Rack-ready checklist (sample)

• [ ] IT kW available at rack PDU within nameplate
• [ ] Grounding and labeling per standard
• [ ] Network demarc patched to row/hall standard
• [ ] Environmental within spec (temp, humidity)
• [ ] Fire/life safety in service for zone
• [ ] DCIM/BMS points live and alarming
• [ ] As-built drawings uploaded
• [ ] Access and safety briefing for install crew

Workload-ready (additional)

• [ ] Racks installed and torqued
• [ ] Structured cabling per design
• [ ] Out-of-band management reachable
• [ ] Cluster/bootstrap smoke test (cluster-deployment-engineer)
• [ ] Change window approved if brownfield hall

Do not declare G5 without G4 sign-off.

Partial delivery

Phased rows acceptable if:

• Safety boundaries respected
• kW accounting clear in burn-up
• Steering informed

Document which racks are in which phase.

Rollback

If commissioning fails:

• Hold G4/G5; root cause owner assigned
• No IT install in affected zone until re-test pass

Reporting and Escalation

Weekly status (delivery program)

1. Headline RAG and demand date confidence 2. Burn-up — kW and racks vs plan 3. Critical path — next 3 milestones and owners 4. RAID top 5 — reds and ambers only in exec summary 5. Decisions needed — numbered with recommendation 6. Budget — spend vs forecast, contingency remaining

RAG rules

Escalation matrix

Steering vs portfolio

Integration with compute demand

When data-center-compute-supply-efficiency shows GPU queue pressure:

• Quantify MW/racks needed vs delivery plan
• Propose accelerate, partial delivery, or cloud bridge
• Do not promise workload-ready without gate sign-off

Anti-patterns

• Reporting green while critical path items are red
• Separate schedules that do not link dependencies
• Declaring rack-ready without signed checklist

Vendor and Construction Management

Vendor roster

Site cadence

• Daily huddle during peak construction (15 min)
• Weekly owner/GC schedule review
• Monthly executive walk (safety + quality)

RFIs and submittals

Aging RFIs on critical path → escalate same day.

Change orders

Log: description, $, schedule days, approver, link to ECO.

No work without approved CO on critical path items.

Safety and access

• Site induction for all visitors
• LOTO procedures for energization events
• After-hours work permits

Quality and punch list

• Punch by zone; target zero critical before G3
• Photo evidence for hidden work before close-up

Acceptance documentation

Package for gate:

• Test reports, warranties, O&M manuals
• Training for facilities operators
• Spares list

Workstream Coordination

Workstreams

Delivery manager integrates schedules; workstream leads own technical decisions.

Weekly integration meeting

Agenda (60 min max):

1. Critical path and float consumed 2. Blockers requiring cross-workstream decision 3. Change orders and schedule impact 4. Safety and site access 5. Next week lookahead

Dependency examples

flowchart LR
  U[Utility energize] --> P[PDU energize]
  P --> C[Commissioning]
  N[Fiber at demarc] --> R[Rack-ready]
  C --> R
  R --> I[IT rack install]

Colo vs owned

Handoff to design lead

Design changes after IFC:

• Impact assessment on date and cost
• Single ECO log; no verbal scope creep