Google's West Virginia deal puts a price on duration

Google has agreed to buy the energy, capacity, and clean energy attributes of a West Virginia project that stacks utility-scale solar with two storage chemistries on a reclaimed coal mine. The Mammoth Solar development in Kanawha County pairs 86MW of PV with 70MW/280MWh of lithium-ion batteries and a 10MW/100MWh zinc-based long-duration storage system from Eos Energy, with output flowing into PJM Interconnection earmarked for Google's data center load in the region, including a planned West Virginia data center.

The timeline makes the project a staged experiment rather than a single ribbon cutting, with the solar array expected to reach completion in 2028, the lithium-ion block in 2029, and the long-duration zinc system in 2030, according to MN8 and Eos. Until that final piece arrives, the site will run as an ordinary solar-plus-lithium plant, and the developers describe the finished build as West Virginia's first long-duration storage installation.

Google has signed up for the project's energy, capacity, and clean energy attributes as a bundle, a structure that lets the developers treat three separate technologies as one dispatchable resource for a customer whose load runs at all hours. MN8 chief executive Jon Yoder said the project reflects his firm's approach to a customer's unique energy profile, combining proven and emerging technologies into one dispatchable resource, and Google's Lucia Tian framed the purchase as backing for round-the-clock clean energy in service of data center demand.

Eos gets a flagship installation out of the deal, one its commercial chief, Nathan Kroeker, describes as deploying American-made long-duration storage where it solves real grid reliability problems. Mammoth Solar is the first project claimed under the master supply agreement Eos signed with MN8 last October, which aims to deploy up to 750MWh of Z3 systems across US renewable projects, making this project the reference point for the rest of the pipeline and the first test of whether zinc chemistry can hold up in live market operations rather than in demonstrations.

The project also extends Google's apparent strategy of buying from technologies that have yet to accumulate a long commercial track record. In recent days, Google agreed to nearly 400MW enhanced-geothermal offtake with Fervo Energy and has a separate partnership with Italian long-duration storage firm Energy Dome whose first project is expected online in 2028. Sell those contracts together and the common thread is duration: Google is buying a load shape rather than cheap electrons, and paying for the chemistry that supplies the last hours of that shape.

PWD has reported that grid access has become the binding constraint on data center growth, and projects reaching completion without an offtake are looking increasingly like merchant risk rather than infrastructure. Mammoth Solar inverts that risk: the offtake exists before the stated completion dates, and the buyer is the load itself. The financing question for long-duration storage has never been whether the chemistry works but whether anyone will pay for a discharge curve that extends beyond four hours. Google has agreed to pay, in the form of a contract that includes capacity, and that is what turns a 2030 completion date into something lenders can model.

The staging carries its own risk: solar and lithium storage begin producing two years before the zinc system arrives, which means the long-duration component will be commissioned into a plant that has already established an operating pattern, putting heavy weight on the 2030 in-service date. A delay in the zinc system would not leave Google without power, but it would leave the long-duration thesis without its evidence base.

The announcement does not disclose what Google will pay, and that gap matters for the rest of the LDES market: without a stated price, it is hard to know what a long-duration megawatt-hour is worth inside PJM. The deal does establish a contract structure other data center developers can copy—one hyperscaler taking energy, capacity, and attributes together and accepting a multi-chemistry stack to get them. Whether that structure appears in the next projects to reach financial close remains unresolved.

The date to watch is 2030. If the Z3 system enters service and dispatches as designed, West Virginia's reclaimed mine will have produced something more useful than electricity: a revenue model for long-duration storage that other developers can carry to lenders, and that model is what Google's procurement arm is commercializing.