Google's Texas battery pilot proves a revenue line, not a price

Texas has spent years giving batteries a market to trade into, but the sharper question is what a battery is worth in the hours that market isn't paying. The newly published results of a three-month Google-sponsored pilot in the state say a second revenue line exists, with one column still blank.

Quintrace's software platform handled verification, LevelTen Energy ran the registry accounts that keep a certificate from being retired twice and wrote the transaction framework for the storage-shift agreements, and esVolta put two of its own Texas projects in: the 240 MW/480 MWh Anole project in Seagoville and the 100 MW/200 MWh Burksol project in Dickens County. Across the three months, those batteries charged and discharged a combined 9.2 GWh in the relevant windows. The instrument was an hourly time-stamped Granular Certificate that allocated solar energy to grid-connected batteries and shifted delivery into hours when Google's consumption ran ahead of its renewable supply.

The structure is the more copyable asset: Google contracted to time-shift environmental attributes it already owned, did not toll the batteries, and took on no merchant or dispatch risk, leaving operators in full control of assets they could still run for merchant price swings and grid emergencies. Set that beside Google's other power transactions and the posture is consistent: it funds PG&E's virtual power plant rather than building generation, it staked the Lea County option before committing capital, it contracted for geothermal capacity whose delivery risk sits with the developer, and it has paid to bring residential batteries into data center service. Google appears 66 times in PWD's records as of September 20, which is what a company looks like when it has become a counterparty to the power market rather than only a customer of clean energy. The constant is that Google buys the attribute or the option and leaves the asset with its owner.

Nine point two gigawatt-hours is worth measuring against the hardware: the two esVolta projects hold 680 MWh between them, so the pilot's throughput comes to somewhere between seven and thirteen full cycles across three months, depending on whether the 9.2 GWh counts each direction of the round trip or the total. Either way, a cycle roughly every week or so, a cadence set by a buyer's load rather than by price spreads, describes an environmental layer that rides on top of a merchant revenue case — a very different thing from a replacement for one.

Tundermann's framing is where the commercial case sits: energy markets pay for certain dispatch outcomes, and the pilot introduces a structure that also pays for the environmental outcome a customer is asking for, a different reason to move a battery than a spread. If that layer generalizes to other hyperscalers carrying the same hourly mismatch between consumption and renewable supply, it becomes a revenue stream worth arguing about; Tundermann's claim that it can support the financing and construction of more batteries is exactly the claim a price would settle. The registry work matters for the same reason: a granular certificate is worth only the confidence that it has not been sold twice, and LevelTen guaranteed exactly that, with the companies describing the pilot as a pathway that retires the hourly certificates behind it.

What the published material does not carry is a price for the storage-shift agreements, a term, or a volume commitment beyond the pilot window — and as this publication has argued about renewable and storage milestones that arrive without an owner, an offtake, or a price, a structure published without a number is a financing flag before it is a market fact. This pilot has more substance than most of that genre, since the certificates were issued, shifted, and retired under a registry, the only thing that makes an environmental attribute sellable at all; but the omitted number still decides what the idea is worth. Ninety days of dispatch data cannot size a revenue line; a price and a tenor can.

Google's restraint is what an operator wants and what a lender cannot yet use: declining to toll means the buyer never holds dispatch risk, so the operator keeps the upside and the exposure in the same hand, and it also means there is no contracted cash flow to lend against. The pathway's value to storage developers therefore rests on someone eventually signing the boring version of the same trade — a priced, term-length agreement for the environmental attribute that a bank will book as revenue. Treating the pilot as contracted cash flow before that happens would be an expensive mistake, because the buyer here kept optionality on both sides of the transaction.

The demonstration itself is real — certificates traveled from solar to grid-connected batteries to a data center's hourly load, with a registry closing the loop at each step — but what has not been demonstrated is a number attached to that movement. Until one publishes, the environmental layer is upside that developers should carry as upside, and the arithmetic that matters is small enough to fit in a sentence: a price, a term, and a counterparty a credit committee will recognize.