Grid access is the new deal currency

Grid access has become the currency of power deals, and Z Squared's agreement to buy Paradox for $25 million shows the structure in pure form: no cash changes hands up front and no debt is raised, while the $20 million earnout pays only if the project reaches energization and lands AI customers. The buyer has acquired an option on power, and the option premium is a contingent payout tied to two milestones that now decide whether a data center project is worth anything at all.

That structure is the new grammar of power deals, and Edisun Power's €470 million binding agreement for Smartenergy operations shows how far the grammar has spread. PWD's deal log recorded that the price values pipeline and grid rights above operating assets, while a conventional valuation of a renewables transaction would start with producing megawatts; this one starts with the right to build, and the premium is explicit.

Sphere 3D is spending on substation capacity for a 50MW Kentucky data center before it has signed a tenant, buying grid capacity and a seat at the zoning table. The 50MW is an option on AI demand, a wager that the scarce input is the connection rather than the building, and the zoning seat matters because local consent is now a condition precedent to energization, rather than a formality.

Greenidge made the same trade at larger scale, repositioning itself around 104MW of energized capacity and a 654MW pipeline in a market where grid access is the driver and the mining rigs and data halls are downstream details. A 654MW pipeline is a queue position that can be sold or financed, rather than a portfolio of finished projects.

Zerra's AU$31 billion Western Downs Digital Park in Queensland sits beside a substation and a ring of gas and solar plants, and the siting is the argument: the developer is betting that grid access, rather than land, sets data center value. A campus without a firm connection is a field; with one, it is collateral that can be financed against, and the substation becomes the anchor tenant of the balance sheet.

Switch's three-building Pittsburgh campus makes the same point from the other direction, pairing 270MW of backup generation with the 345kV lines already on the site. The transmission is inherited, the generation is added, and the combination turns a race track site into a hedge against grid congestion, because the existing lines are the scarce legacy asset and the backup generation is insurance on top of them.

Avaada's $1 billion Haryana campus is selling manufacturers a bundled package of land and generation, and the bundle is the product; whether lenders price it as infrastructure or real estate will decide what it can finance. For the buyer, the value is the certainty of power attached to the land, a certainty that now commands its own line item in industrial site selection.

Microsoft is moving earth on a 469.5-acre Finnish site in Vaasa while permits are still pending, and pre-construction spend before consent shows the binding constraint has shifted from capital to permission. The company is spending to hold its place in the queue, because a delayed permit is the only cost that cannot be refinanced — the earthmoving is the option premium.

Naturgy's 16.2MW solar addition to an existing Spanish wind farm is a grid play in miniature: one interconnection now carries more generation, the cheapest capacity increment available in a constrained queue. That explains why hybrid permits have become so valuable, because the developer adds neither land nor a new queue position; it increases the density of a position it already holds.

Nscale's planned $3 billion US IPO aims to put a $51 billion contracted-revenue backlog in front of public investors while it funds a 10GW buildout, but the backlog is a financial asset only if the grid connections behind it materialize. The IPO is therefore a test of whether public markets will price pipeline the way private buyers now do, and the answer will determine whether the grid option has a public bid.

In Ohio, data centers paid $18 million in grants, a sum that is a rounding error against grid costs but buys something the 85% take-or-pay tariff cannot: community toleration. When a project's biggest external risk is local opposition rather than electricity price, the price of patience becomes a real line item, and the grants are the cheapest insurance a developer can buy against a permit challenge.

Magnora's revised capacity at its Finnish campus in Hämeenlinna makes the same point: the scarce asset in the Nordic buildout is a firm grid connection, rather than land, power purchase agreements, or capital. Every revision of capacity is a repricing of the grid position, and the market is watching that price move because a reduced connection is a permanent impairment, rather than a temporary delay.

The capital is abundant — Cube Green Energy announced a €150 million raise with no terms, and Amprion put €2 billion into transmission in the first half. The queue is not. Developers are therefore paying for pipeline above operating assets, funding substations without tenants, and moving earth before permits, because the one thing money cannot create is a faster interconnection.

That is the correct trade if the grid queue lengthens faster than projects can be built, and the evidence in these deals is that buyers believe it will, structuring their exposure accordingly: paying a premium for pipeline above operating assets is a hedge against a future in which the energized asset is the only asset.

Z Squared's no-cash structure is the purest expression of the trade: the buyer commits no capital today and takes no debt, while the seller's $20 million only arrives if the grid delivers power and an AI tenant signs. That alignment of payout with milestones is a more honest price than any amount of cash paid for a half-permitted site.

The $18 million Ohio patience payments are the smallest number in this story and the clearest: a data center can buy a turbine, lease a building, and hire contractors, but it cannot buy a neighbor's consent. The industry has found a way to price consent, and the next test is whether public markets will price grid access the same way when Nscale's $3 billion IPO asks them to value a $51 billion backlog.