Google puts a price on duration

Google’s solar-plus-storage offtake on a reclaimed West Virginia coal mine buys the ability to store hours and deliver on demand, the difference between an energy asset and a capacity asset—a distinction missing from almost every storage and solar closing this cycle and one whose arrival splits the energy-transition market in two.

Google’s willingness to sign a physical contract there, rather than simply buying renewable energy certificates from a portfolio, suggests the company is underwriting an asset with dispatchable hours—the point of long-duration storage. The counterparties include MN8 Energy and Eos Energy, whose separate 86-megawatt solar-plus-storage pact with Google, reported in this publication’s deal log, tests the same question at smaller scale: whether solar and storage can be sold to a hyperscaler as one firm-power product rather than two bundled renewable credits.

The Fervo-Google geothermal tie-up points the same direction: analysts read it as evidence of baseload scarcity, and William Blair’s risk list locates the doubt in the delivery chain rather than the resource. Google’s demand is for a specific electron at a specific hour, and it is willing to sign contracts with storage, geothermal, and hybrid developers to get it—a preference the West Virginia offtake expresses most clearly by purchasing duration over generation.

New York’s nuclear push after shuttering Indian Point shows the same scarcity from a state’s side: Albany’s carbon-free mandate lost its largest generator, and the state’s new nuclear program is that gap made visible. It tracks the same demand curve as a hyperscaler contract—firm, dispatchable, carbon-free power is scarce, and anyone who needs it is now trying to contract for it directly.

While Google and its peers anchor one end of the market, the rest of the project finance pipeline is still closing without a named buyer or a disclosed price. Equinor’s 100-megawatt, 200-megawatt-hour Texas battery is live, but the revenue contract that would make it infrastructure is absent; Blacktail and RayGen announced a Texas hybrid and named partners and a state, but no capacity, buyer, or price. Alight’s 101-megawatt-peak Finnish solar park has Autoliv attached as a corporate counterparty, but the revenue contract has no named buyer; CVC paid $475 million for the newly completed Gabriela project in Chile while keeping the revenue terms behind that price hidden; and OCI Energy and Arava Power broke ground on 347 megawatts in Texas with no PPA on the record. OroCarbo reserved Haffner technology for US hydrogen projects it has not named, without a project, an offtaker, or an interconnection.

None of these is a failure: a construction event is real, and a turbine in the ground with no PPA is still a turbine in the ground. But it is not yet an infrastructure asset, because infrastructure is a contract that makes a future stream of cash flows predictable enough to finance, and the distinction now tracks line by line—deals with Google or another hyperscaler get duration, price, and a named buyer, while deals without one get steel, panels, and a queue position.

Grid access is becoming an underwriting variable set by customer class and budget line rather than queue position, and a hyperscaler with a signed offtake becomes a customer the utility and the grid operator can name, which changes the schedule. The West Virginia project likely benefited from that logic: a signed Google contract gives the project a reason to be prioritized that a merchant battery without a buyer cannot match.

The split has consequences for who gets built and who gets financed: a hyperscaler’s balance sheet can make a long-duration battery bankable before it is operational because the contract itself is the collateral, while unpriced projects are still being financed like real estate development—build first, hope the market arrives. Merchant storage in Texas has made money in scarcity events and some owners will continue to take that risk, but the Google West Virginia deal shows there is now a premium market for duration-backed firm power, and it is available only to projects with a named hyperscaler offtaker.

The month’s energy-transition activity can now be read as two markets separated by one question: can the project name a buyer willing to pay for time? The West Virginia solar-plus-storage project can; the 347 megawatts in Texas, the Finnish solar park, the Chilean acquisition, and the hydrogen reservation cannot yet. The next test is the MN8-Eos repeat. The first 86 megawatts mattered mostly as proof of concept; a larger follow-on with a disclosed price and duration would confirm that solar-plus-storage has become a product hyperscalers buy rather than a project developers hope they will.