Energization date: when a data center's power goes firm

Energization — spelled *energisation* in British usage — is the point at which a facility can draw firm power at a defined capacity. It is a narrower event than it sounds, and the imprecision is where disputes start.

It is not the same as being connected — the physical interconnection can be complete while the capacity behind it is not yet available. It is not the same as commissioning, which is the testing of the facility's own systems and typically follows first power. And it is not the same as full capacity: most large projects energize in phases, so "energized" without a megawatt figure and a date attached describes almost nothing.

A usable definition has three parts: how many megawatts, firm on what basis, from when. A structure that references energization without fixing all three has left its most load-bearing term open.

The reason the date carries so much weight is that it is the moment the project stops consuming capital and starts producing cash flow. Before it, every layer of the stack is spending — land carry, construction draw, equipment progress payments, standing charges. After it, the facility can accept load, the tenant can take occupancy, the compute can be installed and contracted, and the revenue that repays all of the above begins. There is no other single date in the project that does that.

The date is not one obligation. It is the reference point for most of them, which is why a change to it is never a change to one thing.

Every project forecasts an energization date and most of them miss it. Structures are therefore written around the miss rather than the forecast, and the allocation of that risk is the substance of the negotiation.

The mechanisms are conventional, and it is worth being precise about which does what:

• Target and longstop dates. The target is the plan. The longstop is the date after which a counterparty may walk, reprice, or terminate. The distance between the two is the real contingency, and a structure with a longstop close behind its target has almost none.
• Delay liquidated damages. A pre-agreed amount payable per unit of delay, usually capped. They compensate; they rarely make a counterparty whole, and they never restore a lost commercial window.
• Commencement mechanics. Whether a lease or an offtake commences on a fixed date or on actual energization decides who eats the gap. Fixed-date commencement puts the delay risk on the facility; energization-linked commencement puts it on the buyer's plans.
• Phasing. Splitting the date into tranches — first capacity, then subsequent blocks — reduces the size of any single miss and lets revenue start against the part that is ready. It also multiplies the number of dates that must be documented.
• Contingency and carry. Someone is funding interest, standing charges and site costs during a delay. Whether that is sized into the facility at the outset or negotiated in the middle of the slip is the difference between a manageable delay and a restructuring.

The reason lenders test the date so hard is that delay risk in a data-center project is correlated across the layers rather than independent of them. The same transformer that delays energization delays the lease, the offtake, and the compute deployment behind it — which is why a single date can move an entire structure.

In most sectors the delivery date is an output of the plan. Here it is usually an input, and the plan is built backwards from it.

A tenant with a model to train has a window, not a preference. Capacity that arrives after the window is worth materially less than capacity inside it, and sometimes worth nothing to that tenant at all. So the date is set commercially first, and the supply strategy — grid interconnection, on-site generation, a bridge into permanent power, or some combination — is then selected as whatever can hit it. That inversion is the whole subject of speed-to-power, and it is why behind-the-meter generation is a scheduling decision before it is an engineering one.

Two consequences follow, and they are the practical point of this page.

A date is only as firm as the least firm thing under it. An energization date resting on a queue position that has not cleared its studies is a forecast of a forecast. The stage of the underlying interconnection queue position, not the developer's schedule, sets the honest confidence interval.

Firm capacity and contracted electricity are different questions. A signed power purchase agreement says something about price and volume. It says nothing about whether power can physically be delivered to the site on the date the structure assumes. Projects conflate the two regularly, and the conflation is only discovered at the point it can no longer be fixed.