Power funds filed at zero while credit took $1.1 billion

Hines Rialto Credit Partners, the real estate credit venture of Hines and Rialto Capital, arrived Sept. 15 with $1.1 billion in assets, the largest infrastructure fund filed this week though it owns no plant, pipe, or server. Stockdale Capital Partners filed $300 million, about a quarter of that, and the remaining launches thin out from there: $40 million for thSLVR, a vehicle tied to Theo, and $25 million for the Water and Nature Fund backed by the Walton Family Foundation.

The mandates meant to own the power plants that feed data centers filed at zero.

Infrastructure investing has been framed as a problem of project selection: which site, which queue position, which tenant. The filings point at something upstream of that. The layer that raised this week lends against assets that already exist; the layer that would build the next plant is still forming, and it cannot close until someone prices the output.

Hines Rialto filed capital to deploy against real estate, a business with an unusually clean diligence path: the collateral exists, the income is contracted or can be, and the underwriting runs against buildings standing today. The clean path is the reason $1.1 billion was the number, and it fits a pattern across digital infrastructure. Serverfarm's $895 million add-on and Vantage's $2 billion borrowing base put the leasing forecast rather than a named tenant at the center of the collateral; debt against assets that exist, or very nearly exist, finds a market quickly.

AVK Capital's filing is the one to sit with, because its strategy is to own the generating assets that data centers have financed for themselves, selling hyperscalers relief from the balance sheet by moving the plant onto an outside owner. The vehicle filed Sept. 15 with zero assets under management, unremarkable for a first raise, but the part that matters sits under the strategy: the power purchase agreement that would make a plant bankable still arrives without a counterparty, a term, or a price. A generation fund with no signed offtake has nothing to close against.

EVelution Energy filed the same day, also at zero. Two power-plant ownership vehicles with no capital in a week when the only sizeable check went to a lender is not a coincidence of the calendar. A credit vehicle can close against a portfolio of loans that exist; a generation vehicle cannot close until someone commits to the output, and counterparties do not commit to output without a price, and no price yet exists at a level that makes an uncontracted plant financeable on its own.

The demand side keeps circling the same constraint. OpenAI hired a power screen for its buildout, a single remote role capped at $385,000, whose job is to sort which power options and developer pitches survive contact with a construction schedule. A company that funds a screening desk expects to be shopping for megawatts in a market where the sellers have to be assembled before they can sell anything.

NERC's chief has argued that the grid's real bottleneck is a price on new supply, and the week's filings are that argument in miniature: the interconnection queue is where projects wait, while a financeable offtake is where they move. Those are different problems with different owners, and the vehicles filed this week are positioned on the waiting side and unequipped for the pricing. NERC's chief wants a shift in the approvals habit that built almost nothing for three decades, and the filings suggest that shift has not reached the fund layer.

Schneider has been standardizing the other end of the timeline, packaging factory-built 2.5MW power modules that compress the most controllable segment of a data center build, but what the standardization leaves alone is the grid connection queue, where the schedule is actually lost. A shorter build does not produce a generator, an offtake, or a counterparty; it produces a faster shell around a power problem that has not moved.

The vehicle this market needs is an equity fund for new power capacity, and every structure the week produced routes around one. AVK's strategy presumes data centers have financed their own plants, but a fund that would take that financing off their books needs the plant under contract first; the contract needs a price, and the price needs counterparties willing to hold merchant risk at scale for years—the same counterparties who have declined to hold it. The circle closes on itself: the generation fund cannot raise without the offtake, the offtake cannot be signed at an unbuilt plant's price, and the plant cannot be priced until a fund exists to own it. The week's filings did not break that circle; they showed where it holds.

A first close on either vehicle requires a counterparty willing to sign a long-term offtake at a price that covers an unbuilt plant's cost of capital, or an equity sponsor willing to fund on merchant exposure and sell the contract later. The week's filings suggest nobody yet occupies either role.

This is a harder problem than the sector tends to admit. If the constraint were the queue, the answer would be patience—hold the position, wait out the study, build. The week's filings put the constraint at the first equity check for generation that has no contract and therefore no price, and no vehicle launched this week was built to write that check. Watch the first close. AVK and EVelution both filed with nothing in them, and a first close at either would mean an offtake counterparty finally put a term on paper. Until one does, the week's capital keeps going where the collateral already stands, and the most recent proof is $1.1 billion of real estate credit filed Sept. 15 under the Hines and Rialto names.