Interconnection is the whole deal.

When a family office is shown a grid-tied energy project — a solar-plus-storage site, a wind portfolio, a gas peaker, and lately a data-centre campus that has to arrange its own power — the model it is handed rests on two dates and one number. The date the project reaches commercial operation. The date the offtake starts to pay. The capital cost of getting there. All three sit downstream of a single question the model tends to treat as already answered: can the project connect to the grid, on the schedule and at the cost it assumes.

Most of the diligence I read spends its effort on the offtake and the sponsor and lets a line in the deck stand in for interconnection. That is the wrong way round. For a grid-tied asset, interconnection is not one item among the diligence items. It is the item the others depend on, and it is the one most likely to be described in language that has stopped being true.

A developer will tell you the project is "in the queue," usually with a position number and a date attached, and the number is meant to sound like standing. It is worth being precise about what it is. A queue position is a place in line for a set of engineering studies. It is the right to be studied, and to be told what it would cost to connect and what upgrades the wider system would need before it could. It is not permission to connect, and it is not a cost. It is closer to an option than to an asset, and like an option it can expire, and its value depends on things the holder does not control.

The studies resolve into two questions the sponsor will often blur. The first is whether the project can inject power onto the network at all — the physical interconnection. The second is whether the system will treat that power as firm, something it plans around and counts on, or as energy it will take when the lines are clear and curtail when they are not. In most of the organised markets those are two different products, Energy Resource and Network Resource interconnection service, and they carry different study scopes, different upgrade obligations, and different costs. A project studied for the cheaper of the two can be described, accurately, as interconnecting — while being exposed to curtailment the revenue model never priced. The distinction rarely survives translation into a marketing book. It has to be read out of the study.

For years the queues were run first-come, first-served. Each project was studied in the order it arrived, against a grid that assumed every project ahead of it would get built. Almost none of them were. Speculative positions piled up, each study contingent on the ones before it, and a single withdrawal near the front could force a cascade of restudies behind it. The line stopped meaning very much.

In 2023 the Federal Energy Regulatory Commission rebuilt the process with Order 2023, moving the country's transmission providers from that serial model to cluster studies, and from first-come to what it called first-ready. Projects are now studied in groups rather than one at a time, and to hold a place a developer has to put up more money and show more commercial progress — site control, larger deposits, readiness that is expensive to fake. Withdrawing carries a penalty, and the transmission providers face their own deadlines and penalties for studies that run late.

The reform was overdue and it is a real improvement. It also means that a project's interconnection cost and date are set inside a group, jointly, against the other projects in the same cluster. A sponsor cannot tell you with certainty what its network-upgrade obligation will be until the cluster resolves, because that number moves as projects around it stay in or drop out. When a developer quotes a firm interconnection cost from an early-stage project, the useful question is which study produced it, when, and what it assumed about everything else in the cluster. A number from a feasibility study and a number from an executed agreement are not the same kind of number, and the gap between them is often the difference between the model and the outcome.

The finding that moves the return is almost never in the deck. It is in the most recent study, and it usually concerns network upgrades — the transmission work the wider grid needs before it can accept the project's power. Those costs are assigned to the projects that trigger them, and the assignment is unstable until it is final. When a project ahead in the cluster withdraws, the upgrades it would have paid for do not disappear; they are reallocated to whoever still needs them. A project can wake up to a materially larger interconnection bill without having changed anything about itself. It helps to hold apart the two kinds of cost a study assigns. The interconnection facilities — the project's own line out to the point of connection — are fixed and belong to the project alone. The network upgrades further out are shared, and it is the shared figure that moves when the cluster changes, which is why a careful read keeps the two apart rather than trusting one blended number.

There is a second study that catches people out. When a project is large enough, or close enough to a seam between systems, it has to be reviewed by the neighbouring grid as well — an affected-system study — and that review runs on its own clock, outside the home queue, and can attach upgrade costs of its own. I have seen a project with a clean home-system result and an affected-system finding that arrived a year later and changed the economics. A diligence process that reads only the home-queue documents will not know to ask.

This is the point in the file where the marketing book and the grid-operator's own letters tend to disagree. The book says shovel-ready and interconnection secured. The most recent restudy, or the reassignment notice, or the affected-system result, says something quieter and more expensive. Both are in the data room, or should be. The work is knowing that the second one exists and reading it against the first.

When interconnection is genuinely advanced there will be an executed agreement — in most markets a Large Generator Interconnection Agreement — and its presence is a real milestone. It is worth being clear about what it is. The agreement fixes the interconnection cost, the upgrades required, and a schedule of milestones the project has to meet, up to the point at which it is allowed to energise. It is a cost and a timetable with conditions attached. It is not a switch that is already on.

The milestones are where the risk lives after signing. The agreement will specify dates for posting security, for beginning and finishing construction, for the commercial operation date, and it will attach consequences to missing them — liquidated damages, the loss of posted security, in some cases the right of the provider to suspend or terminate. It will also usually give the developer a right to suspend the work for a period, which preserves the position but resets the clock, and a suspension buried in a status letter is easy to miss and important to find. A signed agreement tells you the project has cleared the studies and priced the upgrades. Whether it will energise on the date the model assumes is a separate question, answered by the milestone schedule and the project's progress against it, not by the signature page.

"Interconnection secured" means different things in different markets, and a reader who carries one market's meaning into another will misprice the risk. In the markets run by PJM, MISO, and the California operator, connection is gated by study and by the network upgrades those studies require; the scarce thing is a firm, affordable path onto the system, and the delay and cost sit in the queue. In the Texas market the philosophy is closer to connect-and-manage: projects reach the grid comparatively quickly, and the system manages the consequences afterwards through curtailment and congestion. Easy connection there is not a free lunch. It relocates the risk from the interconnection cost to the basis — the gap between the price where the project sells and the hub price the offtake references — and a project that connects quickly into a congested pocket can be reliably interconnected and reliably curtailed at the same time.

None of this is a reason to prefer one market. It is a reason to read the interconnection story in the currency of the market the project sits in. In a study-gated market the questions are about the cluster, the upgrades, and the milestones. In a connect-and-manage market the same diligence energy goes into curtailment history, congestion, and the basis risk the offtake does or does not absorb. Applying one market's checklist to the other is how a process misses the thing that actually decides the return.

The documents that carry the answer are not the ones a data room leads with. We ask for the interconnection request and its position, every study in the sequence and not only the latest — the feasibility, the system-impact, the facilities study, any restudy or reassessment, any affected-system result — and the executed agreement with its full milestone schedule and every amendment and suspension. Where the offtake sells firm capacity rather than energy, we ask for the deliverability study that is supposed to stand behind that claim, since capacity a project cannot deliver is capacity it will not be paid for. We ask for the correspondence, because the reassignment notice and the schedule slip usually arrive as letters, not as sections in the memorandum.

Then the work is comparison, which is the part generic tooling does not do. The interconnection cost in the model against the cost in the most recent study. The commercial operation date in the model against the milestone schedule in the agreement. The capacity the sponsor announces against the service it was actually studied for. The offtake's delivery point against the point where the project interconnects, which are not always the same node and whose difference is basis. A summary of each document read on its own tells you little. The finding is in the contradiction between two of them, and it is worth the diligence precisely because it is the thing a busy reader, handed a confident deck, would take on trust.

For a grid-tied project this is not a section of the diligence. It is most of it. The offtake can be investment grade and the sponsor can be excellent, and the return can still be set by a restudy the model never saw. Interconnection is the whole deal, and it reads as the whole deal only if you go and get the study and read it against the story.

— A.O.