Reading a Point of Interconnection

A point of interconnection is a physical place and a legal claim at the same time. Physically, it is the spot where the site's equipment meets the grid — a bay in a substation, a tap on a transmission line, a metering point. Legally, it is a set of numbers: firm megawatts, voltage class, fault duty, and an energization date. Learn to read those numbers and you can price a site's power future without visiting it.

The first fork is transmission versus distribution. A transmission-level interconnection ties directly into the high-voltage backbone — think 115 kilovolts, 230 kilovolts, or higher. It can deliver hundreds of megawatts and it bypasses the congested medium-voltage network, but it requires the site to build and own a substation to step that voltage down, which is capital and lead time the developer carries. A distribution-level interconnection taps the utility's existing medium-voltage feeders — commonly in the range of tens of kilovolts. It is faster and cheaper to connect, but the feeders were never sized for a single load pulling tens or hundreds of megawatts, so distribution connections hit a capacity ceiling quickly. Above roughly a few tens of megawatts, serious compute sites almost always go to transmission.

Voltage class is not a detail — it is a commitment about everything downstream. Connect at 230 kilovolts and you own the step-down transformers, the high-voltage breakers, the protection relays, and the land for a switchyard. Connect at 34.5 kilovolts and the utility owns more of that chain, but you have accepted a lower ceiling. The voltage at the gate propagates through the entire site design: it sets how many transformation stages sit between the grid and the rack, and each stage is a place where capital, losses, and lead time accumulate.

Firm capacity is the number that matters most, and it is easy to misread. Firm capacity is the megawatts the grid guarantees to deliver at all times, including during contingencies when a line or transformer is out. Non-firm or interruptible capacity is cheaper and faster but can be curtailed when the grid is stressed — which, for a data center, tends to be exactly when everyone else is also drawing hard. A point of interconnection advertised at 300 megawatts of non-firm capacity is a very different asset from 300 megawatts firm, and the gap is often the difference between a bankable site and a speculative one.

The energization date is the promise, and it is layered. There is the date the interconnection agreement targets, the date the required network upgrades actually finish, and the date the utility will energize the equipment. These rarely coincide. A transformer with a multi-year manufacturing lead time can sit on the critical path behind everything else; a single large power transformer ordered late can push energization out further than any study. Reading a point of interconnection means reading which of these dates is the real one, and what single component or upgrade governs it.

Then there is the cost allocation buried in the facilities study. Network upgrades triggered by your load may be assigned entirely to you, shared across a cluster, or partly socialized to the broader ratebase depending on the jurisdiction and the tariff. The difference can be tens of millions of dollars and it lands on your project economics, not the utility's. Two sites with identical firm capacity and identical dates can have wildly different returns purely because one sits behind cheap, already-planned upgrades and the other must fund a new transmission line by itself.

Put it together and a point of interconnection reads as a compact risk profile. Transmission or distribution tells you the ceiling and who owns the step-down. Voltage class tells you the shape of the downstream stack. Firm versus non-firm tells you whether the megawatts survive a bad grid day. The energization date, and the long-lead component behind it, tells you when revenue can start. And the cost allocation tells you what the connection actually costs to make real. Everything the rest of this course describes — substations, transformers, distribution, the rack, the chip — is the engineering that turns this legal-physical claim into delivered watts.