Energy and infrastructure.

A queue position is a place in line for a set of studies, not a right to connect and not a cost. We read where the project sits, which studies have actually run — feasibility, system-impact, facilities, and any restudy — and whether it was studied for energy service or for firm network service, which are different products with different upgrade obligations. We separate the project’s own interconnection facilities, which are fixed, from the shared network upgrades, whose cost reassigns when other projects in the cluster withdraw. The document that carries the answer is rarely the deck; it is the most recent grid-operator letter, and the work is reading it against the schedule the model assumes.

The offtake is where the revenue is either real or assumed. We read the tenor against the debt and the useful life, the price mechanism for what it does in the scenarios the model does not show — a fixed price, an indexed one, a hub reference that leaves the project carrying the basis to its own delivery point. We read the curtailment terms for who bears lost output when the grid cannot take it, and the offtaker’s credit for whether the counterparty will still be investment grade in the year the contract matters most. A signed agreement with a weak counterparty and an uncapped basis is a different asset from the one the summary describes.

Construction and connection are governed by two contracts that have to be read together. In the EPC we read the milestone and liquidated-damages structure, the caps, and where cost-overrun and force-majeure risk actually sit. In the interconnection agreement we read the same milestones from the grid’s side — the schedule to energisation, the security to be posted, and the allocation of network-upgrade cost, including who absorbs it if an upgrade is reassigned. The two schedules are supposed to agree. When they do not, the gap is the risk, and it is usually the interconnection schedule, not the EPC, that sets the date the project can actually earn.

A battery is not a fixed asset; it is a depreciating one with a maintenance schedule the model has to carry. We read the degradation curve the warranty guarantees against the throughput the revenue case assumes, the augmentation schedule for when capital has to be spent to hold contracted capacity, and the warranty terms for what voids them. The common error is a revenue model that cycles the battery harder than the warranty allows, so the augmentation the operator will actually owe arrives earlier and larger than the plan. The question is whether the contracted capacity is held across the life of the offtake, and at what cost.

For a data-centre asset the binding constraint has moved from space to power, and the diligence follows it. We read contracted capacity against available capacity — what the utility has agreed to deliver, and by when, against what the site actually needs at full load — and we read the redundancy commitment for whether it is designed or merely described. Efficiency figures such as PUE matter only against a contracted power position that stands behind them. Siting sits underneath all of it: the entitlements, the water for cooling, and the same interconnection question every other project on this page turns on. A campus with a signed tenant and no firm power is a lease against a building that cannot yet run.

Small modular reactors have entered the infrastructure conversation ahead of any settled diligence standard for them. Here the workflow reads the licensing posture — where a project stands with the Nuclear Regulatory Commission, which pathway it is on, and what remains before it — alongside the third-party construction-cost review and the offtake that is meant to underwrite a very long build. This is the one area where the output is explicitly a draft: an orientation for the committee and its counsel, marked not advice, and routed for human review the way the tax brief is, never returned as a regulatory conclusion.