Data Center Commissioning Readiness: What to Prove

Commissioning is where a data center stops being a construction project and starts proving that it can carry critical IT load. That is why data center commissioning readiness should be treated as a commercial and investment milestone, not a technical ceremony at the end of the programme.

I often see delivery schedules showing “commissioning” as a short bar immediately before ready-for-service. The label is neat. Reality is not. Commissioning depends on approved sequences, energised systems, complete controls, trained operators, available load banks, resolved safety constraints and evidence that failure modes have been tested rather than assumed.

The factual baseline is well established. ASHRAE/IES Standard 202-2024 describes commissioning as a structured process with defined roles, documentation and training requirements. The same ASHRAE page lists Guideline 0-2019, which carries the process from pre-design through occupancy and operation. The important point for a data center investor is simple: commissioning begins long before the first integrated systems test.

Data center commissioning readiness starts with the owner’s requirements

The first control document is not the test script. It is the owner’s project requirements, translated into measurable acceptance criteria.

For a critical facility, those criteria should state how much load each phase must support, which redundancy topology is expected, what constitutes an acceptable transfer, how controls should respond to a fault, which operating modes are permitted and who has authority to accept residual defects. If those points remain ambiguous, later testing may be technically successful while still failing the commercial requirement.

This matters when comparing projects in PowerlandMap’s market intelligence. A campus described as “under commissioning” may be testing a single electrical path, a complete building, an initial data hall or an entire phase. Those are not equivalent delivery states.

My view is that every claimed commissioning milestone should answer three questions: what scope was tested, under which operating condition, and against which acceptance criteria? Without those answers, the milestone is narrative rather than evidence.

The readiness gate comes before integrated systems testing

Integrated systems testing is the visible climax, but it should not be the point at which basic completion problems are discovered.

Before that stage, teams normally need signed pre-functional checklists, equipment start-up records, calibrated instruments, complete protection settings, tested controls points, stable BMS and EPMS communications, approved method statements, current single-line diagrams and safe access to every system included in the test.

The sequence is important. Component testing proves that individual assets can operate. Functional performance testing proves that systems respond to commands and abnormal conditions. Integrated systems testing then checks interactions across power, cooling, controls, life safety and operating procedures.

A generator that starts is not the same as a resilient electrical system. A chiller that reaches setpoint is not the same as a cooling train that survives a control failure or loss of utility supply. The distinction mirrors the wider evidence hierarchy described in our guide to data center development readiness: a completed activity only becomes valuable when its scope and dependencies are clear.

Five evidence packages I would request

1. A controlled commissioning plan

The plan should identify systems, test boundaries, responsibilities, prerequisites, hold points, witness requirements and the sequence between energisation, functional testing and integrated testing. It should also show how the programme protects completed equipment from construction activity still under way.

A schedule without prerequisites is optimistic by design. I would expect each major test to be linked to physical completion, controls readiness, documentation and specialist attendance.

2. A traceable issues log

Not every defect blocks ready-for-service. But every open item needs an owner, severity, target date and acceptance route.

The most useful logs distinguish between cosmetic defects, maintainability concerns, functional failures, safety issues and conditions that invalidate a resilience test. They also record whether a workaround is temporary, operationally acceptable or incompatible with the customer’s redundancy expectation.

3. Test scripts tied to real failure modes

Scripts should not merely repeat vendor demonstrations. They should reflect the operating philosophy of the facility.

Typical scenarios may include loss of utility, generator failure to start, UPS module failure, breaker failure, controls-network interruption, cooling plant loss, sensor failure and restoration to normal operation. The exact suite depends on topology and phase. Reported capacity must also match the tested configuration: temporary load-bank arrangements do not automatically prove the full future IT load.

4. Metering and performance evidence

Power, energy, temperature, flow and water measurements should be reliable enough to support both operations and external reporting.

This is becoming more important as efficiency evidence moves from voluntary reporting into regulation. EU Delegated Regulation 2024/1364, adopted on 14 March 2024 and published on 17 May 2024, establishes a common data-center reporting framework for energy and sustainability information. Commissioning is the moment to verify meter coverage, naming conventions, data quality and ownership before gaps become embedded in operations.

The IEA’s Energy and AI report, published on 10 April 2025, also makes clear why electricity demand and infrastructure performance now matter at system scale. That wider context does not replace project testing, but it raises the cost of weak evidence.

5. An operating handover that works in practice

Documentation alone is not handover. Operators need training, access rights, current procedures, spares, vendor contacts, warranty information and enough time to rehearse the facility before customer load arrives.

The NASA Systems Engineering Handbook’s commissioning lesson, updated 1 November 2021, makes a broadly applicable point: commissioning is labour-intensive and requires knowledgeable operators and subsystem experts to be planned in advance. A data center may use different systems, but the delivery lesson is the same. Specialist attendance cannot be improvised after a failed test.

What investors and customers should not accept

I would be cautious when “commissioning complete” is supported only by a certificate, a percentage or a target date.

A percentage can hide uneven progress. Electrical systems may be advanced while controls integration, cooling resilience or operating documentation remains behind. A certificate may cover equipment start-up rather than integrated performance. A target date may depend on utility energisation, authority approvals or vendor attendance that is not yet secured.

This is also why backup generation should be assessed as a system, not a procurement line. Our analysis of data center backup power risk explains how equipment availability, fuel strategy and delivery interfaces can affect the programme long before testing.

The same discipline applies upstream. A supposedly powered site should first pass powered-land due diligence. Commissioning cannot compensate for an uncertain connection, an incomplete permit basis or a phase that has been marketed beyond the capacity actually contracted.

How PowerlandMap treats commissioning status

PowerlandMap separates reported project stage from the evidence behind it. We look for dated announcements, energisation milestones, construction status, stated ready-for-service dates and the capacity basis attached to each claim. Where the public record does not establish the tested scope, we do not silently promote announced megawatts into deliverable IT capacity.

That approach is visible across our market coverage and broader product workflow. It is designed to help investors, developers and users compare projects without flattening every delivery claim into a single “ready” label.

My inference is that commissioning evidence will become more commercially valuable as AI campuses grow larger and phased delivery becomes more common. The first energised building may carry different dependencies from the ultimate campus. A clear evidence chain allows the market to value what is genuinely available today while tracking what remains conditional.

A practical conclusion on data center commissioning readiness

Good data center commissioning readiness is not defined by how close a project appears to its target date. It is defined by whether prerequisites, scripts, witnesses, instruments, operators and acceptance criteria are in place to produce repeatable evidence.

For diligence, I would request the commissioning plan, current issues log, sample scripts, test matrix, metering schedule, operating handover plan and a precise statement of the load and topology being accepted. Those documents reveal far more than a progress percentage.

PowerlandMap helps connect those project milestones with site, grid, capacity and market evidence. If you are assessing a development pipeline or comparing delivery claims across markets, request access to review the platform.

*Matthieu Gallego*