The matching layer in energy infrastructure is the structured intermediation between the three roles required to close any infrastructure transaction — the technology owner, the site or offtake holder, and the capital provider — that qualifies counterparties, structures the deal to a repeatable shape, and moves it to financial close on a defined clock.
It is the layer of work that, in any deal that closes, someone has done. In most deals that don't close, no one has done it.
The three roles it connects
Every energy infrastructure transaction requires three roles, regardless of the technology, geography, or capital structure.
The technology role owns the megawatts — the equipment, the operating data, the technology guarantees, the deployment capability. Examples: a long-duration storage company with eight months of operating performance data, a gas peaker manufacturer with a behind-the-meter product, a fuel cell vendor with grid-scale references.
The site role owns the offtake pathway — site control, interconnection studies or behind-the-meter optionality, regulatory clearance, the offtaker relationship. Examples: a US developer with three 200 MW behind-the-meter sites under option, a West African IPP with a sovereign offtake mandate, a New Zealand site with co-located industrial load.
The capital role owns the money — equity, infrastructure debt, development finance, hybrid structures — and the mandate that defines what it can deploy against. Examples: a $400 million infrastructure debt fund with a US-only mandate, a DFI with a 12-month deployment window in sub-Saharan Africa, a strategic balance sheet looking for vertically integrated positions.
The matching layer is what assembles the three of them around a single closeable structure.
What it does
A matching layer does three specific things, in sequence:
It qualifies counterparties before lawyer time begins. Not a soft introduction. A structured pre-flight: does the technology have bankable performance data; does the site have a credible offtake or interconnection path; does the capital have a mandate that fits the deal's geography, technology, and tenor. Two-tier NDA structure so the conversation can happen before either side spends materially.
It structures the deal to a repeatable shape. A spine that defines the offtake price band, curtailment assumptions, credit support, dispatch logic, technology guarantees, site control, and financing waterfall. Bespoke where bespoke matters; standardised where standardisation makes the deal close.
It moves the deal through documentation and close on a clock. Term sheet inside four weeks of qualification. Definitive documents inside sixteen. If any side falls off, fail fast and re-match — better a dead deal in week eight than a dragged deal in month nine.
What it is not
The matching layer is not a marketplace. A PPA marketplace lists prices — useful for understanding the curve. The matching layer assembles a deal between three counterparties who must negotiate around each other simultaneously, not bid on a price.
It is not brokerage. A broker introduces and earns on a transaction; the matching layer qualifies, structures, and earns on a closed deal at a much lower attrition rate.
It is not advisory. A sell-side advisor represents one side of the table. The matching layer sits structurally in the middle, with no positional preference for which side wins.
It is not procurement. Corporate procurement teams run an internal mandate against an external market; the matching layer connects external mandate to external mandate to external mandate.
The closest analogue in another industry is buy-side M&A advisory — where the work is constructing a deal that didn't exist before, not selecting from deals that already do.
Why it matters now
The matching layer wasn't necessary at scale until roughly 2023, when the AI workload reset transaction volume, complexity, and geography in eighteen months. US data centre power demand climbed from about 50 GW in 2024 toward 76 GW in 2026. Hyperscalers built the matching capability internally — bilaterally, deal by deal, by pulling 17-year operating-asset PPAs from existing nuclear plants. Below hyperscaler scale, no balance sheet does this in-house. The layer has to be a service.
The deal-mortality consequence: across the broader US energy infrastructure market in 2025-26, the base rate of structured deals that fall over after term sheet sits at 60-70%. A matching layer that pre-qualifies on credit support, deployment window, regulatory exposure, and mandate alignment plausibly cuts that to 30-40% — not by making deals easier, but by killing the wrong deals earlier.