Commercial operation date: the milestone that starts rent

The commercial operation date — COD — is the date on which a facility is treated by its contracts as ready to perform its commercial function. It is defined in the documents rather than observed in the world: a set of tests, a set of conditions, a certificate, and a date.

The three parts of a workable definition are always the same, and a structure missing any of them has left its most consequential milestone open:

• What has to be true. Capacity available at a stated level, redundancy demonstrated, systems accepted, permits and operating approvals in place, agreed exceptions listed.
• Who decides that it is true. The independent engineer, the tenant, the offtaker, the lender's technical adviser, or some combination — and what happens when they disagree.
• What issues on the date. A certificate, a notice, or a deemed occurrence. Whichever it is, it is the thing the obligations key off.

The reason COD is drafted so carefully while energization often is not is that COD is where the money changes direction. Before it the project is drawing; from it the project is paying and being paid. A milestone that turns cash flow around is a milestone both sides will litigate over the definition of, and the definition is cheap to fix at signing and expensive to argue about afterwards.

One clarification worth making early, because it is the most common misuse of the term: COD is a contract term, not a status of the building. A facility can be complete, energized and physically capable of running while COD has not occurred, because a condition in the definition has not been satisfied. It can also be deemed to have occurred while the facility is not yet running, because a party caused the delay and the documents said so.

The two milestones get used interchangeably in marketing materials and almost never in documents, which is a reliable signal about which audience each was written for.

Energization is the availability of firm power to the facility at a defined capacity. It is an input. COD is acceptance of the facility as ready to perform, and it is an output — of energization plus everything that has to happen once power is present. Between the two sit commissioning, testing, integration and whatever remediation those produce, and that interval is real, planned for, and routinely compressed to nothing in a schedule shown to a buyer.

The gap is also why a long grid timeline is quoted the way it is. the typical time from an interconnection request to commercial operation now runs beyond five years, against a median of under two years in the mid-2000s, on a US queue holding well over a thousand gigawatts of proposed capacity (Lawrence Berkeley National Laboratory, as of August 11, 2026) — a figure quoted to commercial operation rather than to first power, because commercial operation is when the asset does anything for anyone.

The milestones below are the ones a data-center structure typically names. They are listed in the order they occur, and the point of the table is the third column: they do not release the same things.

The practical value of separating the milestones is that each one is the trigger for a different instrument, and the instruments were negotiated by different parties who did not necessarily use the same word.

The construction facility is sized to a period, and the period ends at a milestone. A construction loan funds against progress and carries interest that nothing is yet earning. The date it stops being a construction facility and starts being longer-term debt is a conversion condition, and conversion conditions are drafted around completion and operation rather than around first power. The mechanics of that conversion — what has to be satisfied, who controls it, what happens when it is not met — are set out in construction-to-permanent for a data center.

The lease starts paying on its own trigger, which may not be COD. Rent commencement is a lease concept and COD is a project concept, and they are aligned only if somebody aligned them. A lease commencing on a fixed date against a project reaching COD on its own schedule has put the delay risk on the landlord; a lease commencing on actual COD has put it on the tenant's plans. Neither is wrong. Both are decisions, and the expensive version is the one nobody made.

The offtake starts on a third trigger. Where capacity is sold under a contract rather than a lease, the payment obligation begins on a date defined in that contract — which may reference the facility's COD, a phase of it, or a date of its own. A structure with three documents referencing three differently defined milestones has a gap in it, and the gap is discovered when the first invoice is disputed.

Equity is last and cares most. Distributions are typically blocked until the facility is operating and coverage is demonstrated, which means the sponsor is carrying everything until the milestone the debt was drafted around. Moving COD by a quarter moves nothing for the contractor, something for the lender, and everything for the equity.

The discipline that follows is unglamorous and it prevents most of the disputes: write one definition, reference it from every document, and make sure the party who tests it is named in all of them. A project with a single defined COD and three documents pointing at it behaves predictably. A project with three definitions has three projects.

Large facilities rarely reach commercial operation once. They reach it in blocks, which turns one negotiated definition into several and introduces the questions that most often go unanswered.

Phased COD. Each phase has its own capacity, its own tests and its own date, and revenue starts against the part that is ready. That is generally the better structure: it de-risks the schedule and starts cash flow earlier. What it also does is create the possibility of a document that refers to "the commercial operation date" without saying which one. Where a facility has four phases, the definite article is a defect.

Deemed COD. Where a party's own act or omission prevents the milestone being achieved — a tenant that will not schedule acceptance testing, an offtaker that cannot take delivery — the documents commonly deem the milestone to have occurred anyway, so that obligations begin. It is the right mechanism and it is frequently omitted. Without it, a counterparty can defer its own payment obligation by declining to participate in the test that would start it.

The longstop. COD carries a backstop date after which a counterparty may terminate, reprice or claim. The distance between the target COD and the longstop is the real contingency in the schedule, and a structure whose longstop sits close behind its target has almost none. The allocation of the cost when the date moves is a separate question, and it is worked through in energization rather than restated here.

And the criteria have to be achievable. A COD definition requiring full design capacity and full redundancy demonstrated simultaneously, on a facility built to be filled in stages, describes a test the project will fail on schedule and pass late. Acceptance criteria that do not match the way the asset was actually built are the quiet cause of a surprising number of missed dates.

The test to apply to any draft definition is short: can a reasonable engineer, standing on the site with the documents, say yes or no today. Where the answer requires a negotiation, the milestone is not defined — it is scheduled.