Naturgy pays €330m for the grid it already holds

Naturgy has agreed to pay €330m in Spain for grid it already holds, an odd purchase only until you count what has become scarce. The repowering programme behind the cheque adds megawatts at connections already in hand, and the coverage is exact about the money and silent on the offtake. Repowering is the cheapest way to put capacity onto a network, but the megawatts are the easy part: who buys the output, at what price and for how long determines whether €330m was a bargain or a rounding error.

The same trade turns up in Australia with the price filed off: Palisade's 1.4 GW pipeline buy was announced without a number and without the detail that would say whether the asset is land or a position in a connection queue. In a market where connection rights set the value of renewables development, that distinction is most of the valuation, and the disclosure stops at the headline.

Naturgy is buying megawatts behind a wire it already owns; Palisade is buying places in a queue somebody else controls. Both are paying for a network that is not being built fast enough to make the question moot, and both at prices the market is only sometimes told. What repriced this cycle is permission to move power — a queue position, an interconnection agreement, a point of delivery that already exists — and this week's announcements are where the comps get set, when they get set at all.

A connection right outlives the project attached to it for a plain reason: a solar farm or a battery can be rebuilt somewhere else, and the wire cannot be moved. Developers have understood that for years, which is why land banking gave way to queue banking; what looks different this cycle is that the buyers of those positions arrive with balance sheets large enough to pay for the connection and let the generation follow it.

The logic reaches even the deals that need no new wire at all. The fastest AI capacity in the market sits inside buildings already standing, and the electrical ceiling there is the lowest in the business; retrofit space is quick because the connection exists, and it stops being quick the moment the substation fills. That is a connection constraint wearing a real estate costume, and it is why the same queue keeps surfacing in transactions that appear to have nothing to do with grids.

Naturgy and Palisade are the sharpest cases rather than the whole pattern: Hy24's purchase of the OPAL gas pipeline names a buyer and a seller and withholds the one figure that would settle the trade, while Latvenergo's pursuit of 250 MW arrives with no seller, no price and no offtake attached. Nordex's 34-MW German repowering order is a capacity number with no buyer, site or money behind it, and the quarter's renewables round-up — Qualitas Energy, Ørsted, Alinta and Sonntag — carries no prices, counterparties or capacities either.

Masdar and Luxcara's EUR5bn tie-up runs the pattern backwards: a headline number and two technologies, with no capacity, counterparty or structure, so the figure floats without anything to test it against. Cox's $800m hybrid refinancing of a Mexican project carries no coupon, tenor or ranking, while a 200-MWh Colorado battery has its first named lender on the record and still no loan amount or offtake. TAR's $120m raise for off-grid AI power names neither counterparty nor instrument.

The data center side has an unpriced input of its own: governance now gates AI throughput at the rack, arriving as the one capital cost tenants have no line item for. It sits inside the largest leasing market in infrastructure with the same shape as a pipeline announced without a number.

The easy explanation is that disclosure has slipped and diligence has got harder; the more useful one is that the absent number is the asset. Once the scarce thing is a queue position, a connection right or a signed offtake, the price at which one changes hands is exactly what the other side of the next negotiation wants to know, and a market that publishes it is handing out comps. Deals go quiet where a connection is what is being sold.

Bannock County, Idaho, is the live price discovery this week: the county shut its door on utility-scale solar and wind and has now reopened the question with a reversal vote, turning a planning decision into a test of whether permitting consent — the binding constraint on American power — can be bought back on better economics. If it can, a local refusal stops behaving like a political constant and starts behaving like an input cost.

The stakes run past one county's agenda, because consent sits upstream of every other number in a development model: a project without it has no connection, and a project without a connection is a spreadsheet. A reversal that clears on improved terms would suggest the refusals of the past few years are renegotiable rather than final, and that is a larger finding for a developer than any single vote: the counties that already said no become the first places worth a second conversation.

Pennsylvania is running the same experiment through a docket rather than a ballot, where a rewrite of curtailment rules for data centers, standing beside a cost-of-capital proceeding, will decide which load in PJM gets financed and at what hurdle rate. One is a vote at the county line, the other a rate case at the commission, and both are pricing the same commodity: the cost of getting onto the network. Commissions elsewhere tend to read the first mover's order closely, and the number Pennsylvania lands on is the one they will benchmark against.

The states have been accumulating this authority for a while: when the D.C. Circuit read section 202(c) narrowly, the retirement calendar went back to them, leaving dispatchable capacity to earn its premium in capacity auctions and state procurement dockets rather than by federal order. Put that beside Pennsylvania's cost-of-capital proceeding and the venue for pricing grid access is shifting to state commissions, with each order a template for the next.

Egypt makes the administrative version of the trade explicit: Heca Data's power study with the state transmission company, and not its September 3 meeting, will decide whether the integrated hyperscale zone becomes a project at all. Where the transmission owner is the state there is no queue in which to buy a position; the connection is granted, and the study is the instrument that grants it. Capital will not move into the zone until the study confirms the watt, which puts a connection right inside a government process instead of a market — a slower, less liquid and considerably cheaper way to hold the same scarcity.

Friesen Elektra is running the purest version of the trade by monetizing power rights rather than building data centers: the same asset with the shell taken off, and also the version most exposed to a grant outliving the deal it was priced off.

Orbital shows how far the logic reaches: a 10 GW constellation holding $5m in the bank and an FCC filing still pending is a licensing wager before it is a hardware one, because capital on that scale follows an approval rather than arriving ahead of it. The venue changes from a substation to a spectrum allocation, and the thing being underwritten stays the same.

There is a second half to connection economics that this week's data center work makes plain: load that can sell grid services — on-site storage, an efficient direct-current backbone — comes to the utility with something to trade, while load that simply asks for power arrives in a queue. Where utilities already refuse the fast-deploy diesel bridge they distrust, that difference decides who gets connected and who waits. The operators who build for flexibility will buy access more cheaply than the ones who built for demand, and the gap between the two will widen as the queue lengthens.

Reliability planning reached the same conclusion from the other direction: peak outages fell sharply after Elliott and Uri, and the credit sits with FERC and NERC coordination and transfers between regions rather than with new steel. Using the wire that already runs is cheaper than building another one, which is the conclusion Naturgy reached in Spain at a much smaller scale.

Watch Bannock County first. A yes would not prove that consent is purchasable everywhere, but it would put a public number on what a refusal costs to reverse; once that number is in a development model, counties sitting on old refusals stop looking like dead ground.