A project can be fully installed and still be unable to deliver the capacity its dashboard promises.
The equipment is present. The model is complete. The handover folder is populated.
Then an operator asks: “Can we put the required load on this system today?”
That question exposes the distance between a number recorded against an asset and a capability demonstrated by the assembled infrastructure.
For owners, that distance can mean delayed revenue, restricted operations, additional testing, or an operating decision made with more confidence than the evidence supports.
It is also where digital twins need to become more useful.
A cooling deployment makes the distinction concrete
Vertiv’s October 16 announcement describes a direct-to-chip liquid-cooling deployment at CapitaLand India Trust’s CLDC Mumbai 01 data center in Navi Mumbai, supporting 30 MW of AI workloads. The company reports technology-loop balancing and Level 4 testing ahead of go-live.
The significance is the connection between physical installation and performance validation.
However, a public announcement is not the underlying commissioning dossier. It does not establish every test condition, acceptance criterion, failure scenario, or continuing operational limit.
We should therefore distinguish the reported deployment from the broader lesson it illustrates: a capacity claim becomes more useful when its supporting evidence and boundaries are visible.
Thirty megawatts also needs a clear definition. Cooling capability, electrical supply, IT load, and useful computing output describe different things. A shared unit does not make them interchangeable.
Capacity needs a scope
Consider an illustrative pump labelled “500 liters per second.”
That figure leaves several questions unanswered. At what head? In which configuration? With what fluid? For how long? With which upstream and downstream constraints?
A manufacturer’s rating, a design calculation, a site test, and current operating telemetry each contribute different evidence.
They should retain those distinctions inside the asset record.
A test result supports a claim within the conditions evaluated. Formal acceptance records an accountable decision against agreed criteria. Continuing availability depends on whether the relevant conditions still hold.
My concern is what happens when those distinctions disappear during digital handover.
A carefully qualified engineering statement can become a single unqualified field: “Capacity = 500 L/s.”
The data survives. The meaning becomes weaker.
Give the number a history
I would use six separate capacity states in an infrastructure information model:
State - Question it answers
Designed - What does the engineering intend to deliver?
Installed - What equipment and configuration are physically present?
Tested - What performance was demonstrated, under which conditions?
Accepted - What scope has the responsible authority accepted?
Deliverable - What can be made available under current constraints?
Utilized - What is actually being used?
This is a proposed management framework, not a universal commissioning standard.
The states need not advance together. Part of an installation may be accepted while another part remains restricted. Maintenance can reduce deliverable capacity without changing the installed equipment count.
Keeping these states separate helps a project director locate the remaining work and an operator understand the usable margin.
It also gives finance a clearer basis for discussing when installed investment becomes available service.
Build a capability evidence record
For each important capacity claim, I would connect three groups of information.
The claim: the asset or system identity, capability, unit, physical boundary, configuration, and requirement being assessed.
The evidence: the test procedure, conditions, measurements, instrument references, observed results, exceptions, and links to source records.
The decision: who reviewed and accepted the result, what restrictions apply, when it took effect, and what changes require reassessment.
The original report should remain accessible. Structured fields make its relevant meaning searchable while preserving the source and its revision.
This matters especially when equipment changes. A replacement pump may occupy the same functional position, but its predecessor’s test history should not silently become evidence for the replacement.
The relationship between physical identity, installed configuration, and evidence needs an owner.
Connect spatial context to engineering evidence
BIM can identify intended systems and their relationships. Reality capture can help check installation and geometry. GIS can locate external dependencies. Sensors can show current behavior.
Each source answers part of the question.
A scan may show that a cooling skid is present. It cannot, by itself, demonstrate thermal performance.
A sensor may show acceptable temperatures during a low-load period. That observation does not establish performance at the claimed peak.
A commissioning report may demonstrate a defined configuration. Subsequent modifications can change whether it remains applicable.
The digital twin should preserve these relationships and expose missing links.
For an owner, a useful view might show installed equipment alongside accepted scope, open exceptions, constrained capacity, and the evidence required for the next release.
Commissioning evidence must remain connected to change
Commissioning provides a baseline. Operations introduces changing conditions.
Equipment is replaced. Control settings change. Dependencies become unavailable. Loads grow. Measurement quality deteriorates.
I would therefore attach reassessment triggers to important claims.
A configuration change could mark the affected claim for review. A failed dependency could restrict the deliverable-capacity assessment. A new test could support a revised acceptance decision.
Those transitions should follow approved engineering and operating rules.
For AI-assisted operations, this becomes a practical boundary. An agent recommending additional load should be able to retrieve the relevant accepted envelope, current restrictions, and unresolved evidence gaps.
Where that information is missing, the system should make the uncertainty visible and route the decision for review.
The commercial opportunity is in the handover
For BSMA, I see an opportunity in helping owners connect asset identity, BIM/GIS context, commissioning records, and operating information.
My partners could provide the integration and decision-workflow layer for that approach, developed with commissioning specialists and operators. Technical testing and acceptance would remain with the responsible domain authorities.
A focused starting point would be one critical system: reconcile its installed state, map its capacity claims, connect the evidence, identify gaps, and define change triggers.
The value should be measured through fewer unresolved handover gaps, faster evidence retrieval, and clearer decisions about available capacity.
An infrastructure dashboard should let an operator distinguish what is present, what has been demonstrated, and what can be relied on today.
The next useful question about capacity is: “Which evidence supports this number and what would make us reconsider it?”
