What is an interconnection queue?

An interconnection queue is the ordered process by which a utility, a transmission owner, or a regional system operator evaluates requests to connect something new to the grid — either new generation feeding in, or, increasingly, new load drawing out. Requests are studied in sequence, each against the assumption that everything ahead of it in the order is built.

A queue position is a place in that sequence. It is not an entitlement to power. It is standing in a process, and it is defined narrowly: a named parcel or point of interconnection, a stated capacity, a stated in-service date, and a set of study assumptions about the network around it. Change any of those materially and the request is generally restudied, which in practice means it moves.

The reason this matters more than it used to is arithmetic. Queues in the major markets are measured in years, while AI demand is underwritten on horizons measured in quarters. When the process is longer than the commercial window, position in the process becomes the binding constraint on the whole project — not land, not equipment, not capital.

The consequence for anyone reading a site: the first question is not how many megawatts are available. It is what has been requested, when, against which parcel, and where that request currently sits.

A queue position has the two properties that create scarcity value, and one that limits it.

It is time that cannot be bought back. Every month in the queue is a month already served. A position established two years ago is worth more than an identical request filed today for the same reason a lease with two years of seasoning is not the same as a new one — the calendar has already been paid for.

It is not fungible. A position is attached to a point on the network. It cannot be lifted and dropped onto a better parcel, and its value depends on grid conditions at that specific point, which the holder does not control. Two positions with identical stated capacity in the same market can be worth entirely different amounts because one sits behind a constrained transmission path and the other does not.

And it is conditional throughout. A position carries obligations — deposits, milestone dates, security postings, study cost payments, and demonstrations of site control and readiness. It survives only as long as those are met. This is the property that most often surprises the holder: the position is not owned, it is maintained.

What follows for capital is that the position is treated as a gating condition rather than as collateral. Very little of a queue position is realisable on its own, because its transferability is a matter of tariff and utility consent, not of the holder's intent. Its function in a structure is to make the layers above it possible — the seam described in [the data-center capital stack](the-data-center-capital-stack).

Every market runs its own sequence under its own names, and the names change. The underlying ladder does not: each rung converts some part of the uncertainty into a number, and only the last one creates an obligation on anyone.

Underwriting a queue position is mostly the study of how it fails, because the failures are procedural rather than commercial and they are not visible in a marketing pack.

• Milestone default. Missed deposits, missed security postings, missed demonstrations of site control. The most common cause, and the most avoidable.
• Restudy. A material change — capacity, parcel boundary, in-service date, point of interconnection — can require the request to be evaluated again, against a network that now contains everything filed since. The position survives on paper and the answer changes.
• Withdrawal cascades ahead of it. Because each request is studied assuming the ones ahead are built, projects dropping out of the queue in front can trigger reassignment of upgrade costs onto those behind. A position can get materially worse without its holder doing anything at all.
• Parcel mismatch. The interconnection was studied against one parcel and the site control covers another, or a boundary moved during assembly. This surfaces in diligence and it is treated as fatal, because it means the studied position does not describe the site being sold. The wider version of this test is set out in the [site financeability gates](/sites/what-makes-a-site-financeable).
• Process change. A position established under a procedure that has since been replaced is a position in a process that no longer works the way the holder assumes. Whether it converts into the successor process, and on what terms, is a specific question with a specific answer — and it is the question behind [verifying a power claim](/sites/verifying-a-power-claim).
• Entitlement failure underneath it. A queue position over land that cannot lawfully host the use is standing in a process for a project that cannot be built. See [entitlement](entitlement).

Each of these is checkable in advance. None of them is checkable from a capacity number.