What do you do with a 100-hour battery?

Kann opens by complaining that “long duration energy storage” has been stretched from six hours to seasons until it means nothing, and the conversation replaces the category with a function. Lithium-ion, Jaramillo says, is essentially a power battery doing intraday work: it started at fifteen-minute frequency response in PJM and now sits at four hours, moving toward six, providing peaking and ramping. That maps onto gas peakers, which the industry defines as plants running less than 5% of the hours in a year, roughly 400 hours, spread out rather than consecutive. There is no technical reason a lithium-ion battery cannot discharge for a hundred hours; the limit is economic, and it turns on rated power. Every additional hour of discharge at rated power is a cost adder, roughly $100 per kilowatt-hour for lithium-ion, and because the industry compares resources on dollars per kilowatt, a hundred hours would mean paying about 25 times as much per kilowatt. That does not clear. He also does not expect anyone to unseat lithium-ion in that intraday band, because grid storage rides on the manufacturing scale of the automotive market and no other chemistry gets the same tailwind; he calls himself a technology optimist and calls displacement a tall task anyway.

The more interesting claim is that value does not rise smoothly with duration. Run a capacity expansion model and ask what a storage asset displaces, then walk up the histogram of gas plant capacity factors from peakers through mid-merit plants at 20, 30, 50, 70%. Form’s modeling, led on the analytics side by co-founder Marco Ferrara, found clear value up to about 10 or 12 hours, then markedly less value until you get back up to around 75 and approaching 100 hours, because that is where a battery can start replacing what those mid-merit plants do. Jaramillo is careful about the middle: he explicitly declines to put a number on how much value sits between 10 and 40 hours, allows there may be a day-to-day shifting case, and says it still needs to be worked out. The reason the answer lands near four days is not total hours but consecutive ones. The signature of severe weather is three to five days, whether that is a Pacific Northwest heat dome, an Upper Midwest polar vortex, a nor’easter or Winter Storm Uri, and a grid increasingly driven by weather has to be able to ride through that.

None of which matters without a price, and Jaramillo is insistent on this: the assumption that multi-day duration is irrelevant is really an assumption about what it costs. Form’s own modeling, which he says third parties have confirmed, put the threshold at 100 hours and $20 per kilowatt-hour. Kann does the comparison out loud, noting lithium-ion cells around $100 per kilowatt-hour and fully installed grid systems at two or three times that, which makes the target roughly an order of magnitude below lithium-ion on an energy basis. The justification for thinking it reachable is what Jaramillo calls entitlement. Iron-air chemistry is old and was never commercialized, but its active materials cost less than a dollar per kilowatt-hour dug out of the ground, against $30 to $35 for lithium-ion’s unprocessed active materials. Starting cheap is necessary but not sufficient, so the engineering constraint is self-imposed: no expensive synthesis and no high-precision manufacturing, because the whole point is to end cheap as well as start cheap.

What utilities buy it for today is not the renewables balancing story. Georgia Power is dealing with load growth fast enough that it reopened its integrated resource plan two years early; multi-day storage lets it add cheap intermittent generation to meet that growth without giving up reliability, and because Georgia Power runs internal markets rather than bidding into a wholesale market, it can put a precise internal value on reliability in a way wholesale markets do not. Jaramillo adds a land argument, that the same benefit can be had from about half as many acres of solar, while noting he does not want to be a land doomer because power projects are a small share of industrial land use. At Xcel, which is wind-heavy and participates in MISO, the same asset also works as a physical hedge against price spikes, something otherwise available only financially or through thermal plants. A third application, buffering constrained transmission lines and letting curtailed wind owners move energy across days to escape the price spread between where they generate and where they get paid, had been shelved early and come back quickly. As for how the battery behaves, Kann offers the simple picture of charging up, waiting for a scarcity event and discharging for three days, and Jaramillo rejects it: that arbitrage is lithium-ion’s game in ERCOT. The justification is capacity and reliability, and then the asset gets used for everything else, running flat out for three or four days a few times a year but otherwise ratcheting its state of charge up across a renewable-heavy spring and down across a summer deficit.