What is behind-the-meter generation?

Behind the meter describes generation sited on the customer's side of the utility's revenue meter, serving the customer's own load directly. Electricity produced there does not pass through the transmission network and is not sold into a market; it is consumed where it is made.

The contrast is with front-of-meter generation, which connects to the grid, sells into it, and reaches a facility as a delivered commodity under a tariff or a supply contract.

Between the two extremes sit the arrangements most large projects actually use:

• Islanded — the facility runs on its own generation with no grid connection at all. Rare at scale, because it makes availability entirely the project's problem.
• Grid-parallel — on-site generation carries the load with a grid connection retained for backup, top-up, or standby service.
• Bridging — on-site generation carries the facility until permanent grid supply arrives, after which its role changes or ends.

The reason the arrangement matters commercially is well covered elsewhere on this site: interconnection queues are long, and a facility that can generate its own power is not waiting in one. [Speed-to-power](speed-to-power) sets out that argument and the tactics behind it. This page assumes the decision has been taken and asks the question that follows it — what changed about who is exposed to what.

Buying power from a utility is a procurement relationship with a regulated counterparty. Generating it is an operating business. The differences are not marginal.

The transfer above is not a reason against behind-the-meter generation. It is the list of things that have to be documented before the structure works, and each one has an established form.

Fuel becomes the mirror image of offtake. A project selling contracted capacity against uncontracted fuel has written an unhedged spread. The instruments are conventional — supply agreements, firm transportation, hedges — and their tenor should be read against the tenor of the revenue they support. Fuel arranged for one year behind capacity contracted for ten is an open position, not a saving.

Availability becomes a credit term. Under a tariff, reliability is the utility's obligation and largely invisible. On site, it is the project's, and it is what the tenant and the lender are relying on. Redundancy, maintenance regime, spares holdings and an operator with a record are underwritten in the same way an operating asset would be, because that is what it now is.

Permits constrain operations, not just construction. Air permits typically limit running hours or emissions over a period. A generation set that is lawful as backup and unlawful as baseload is a common and expensive discovery, and it is an [entitlement](entitlement) question rather than an equipment question.

The grid connection usually still exists, and still has to be arranged. Standby and backup service is a tariff product, priced, and often subject to its own queue. A facility that is islanded in normal operation and grid-connected in contingency is holding two positions, and the second one is frequently discovered late. On-site generation shortens the critical path; it rarely removes the [interconnection queue](interconnection-queue) from it entirely.

And the generation is a separate asset. Turbines, switchgear and substation equipment have their own economic life — long, and quite unlike the compute they power. Once they are on site rather than on a utility's balance sheet, they are financeable on their own terms, by equipment finance and leasing capital that will not touch the layers around them. That separation is the point of [the data-center capital stack](the-data-center-capital-stack), and behind-the-meter generation is the layer where it becomes unavoidable.

The single most consequential thing to establish about behind-the-meter generation is how long it is meant to matter.

Permanent on-site generation is the facility's primary supply for the life of the building. Its equipment is a twenty-year asset supporting a twenty-year role, and it can be financed on that basis with a matched tenor and a residual that means something.

Bridging generation exists to carry the facility until grid supply arrives. Its role ends on a date that is itself uncertain, and the honest structure says so. Financing a bridge as though it were permanent produces an asset with a long facility against it and no job to do; financing it too short forces a refinancing at the worst moment, when the grid date has slipped and the units are still needed.

The test that resolves it is redeployability. Equipment that can be moved, resold or re-contracted when its role ends is carrying a genuine residual, and its exit does not depend on the project succeeding. Equipment specified so tightly to one site that removing it costs more than it fetches has no residual at all, whatever the schedule says — which is the mechanism set out in [stranded assets](stranded-asset).

The corollary is worth stating plainly, because it decides more structures than any technical consideration: the right answer is rarely all-grid or all-onsite. It is the supply mix that hits the [energization](energization) date at a cost and a risk profile the capital behind each layer will accept — and the mix is a structuring decision made against contracts and tenor, not a preference about equipment.