Beyond Lithium: The Race for Long-Duration Storage and Why the Next Decade Depends on It

Lithium-ion batteries have done something extraordinary in the past five years: they have made the gas peaker, that long-unloved workhorse of the modern grid, economically obsolete in a growing number of markets. A four-hour lithium-iron-phosphate battery, charged on midday solar electrons, can now discharge into the evening peak at a delivered cost that undercuts the marginal cost of running an existing combustion turbine; let alone the cost of building a new one. This is, by any measure, a historic achievement. It is also, paradoxically, the easy part.

The hard part is what comes next. A four-hour battery can shave the evening peak. It cannot get a regional grid through three consecutive cloudy, windless days in January. It cannot bridge a multi-week winter cold snap that drops temperatures across half a continent. It cannot replace the seasonal storage that natural gas, in its capacity as a fuel that can be stockpiled in vast underground reservoirs, currently provides almost invisibly to every modern power system. The four-hour battery is a beautiful solution to one specific problem. It is not a solution to the larger problem of how to run a grid that is 80 or 90 percent renewable.

The larger problem requires what the industry calls long-duration energy storage, or LDES; systems that can discharge at full power for ten, fifty, one hundred, or even five hundred hours. Whether the energy transition succeeds at the pace and scale the climate requires depends, more than on any other single technology, on whether LDES becomes economically viable at scale within the next decade. It is the most important and most under-discussed technology race in energy.

The economics of lithium-ion get progressively worse as duration extends. A four-hour lithium-iron-phosphate battery costs roughly $245 per kilowatt-hour of energy capacity installed. An eight-hour battery costs roughly $350. A twelve-hour battery costs roughly $500. The reason is structural: in lithium chemistry, the cells that store energy are the same cells that deliver power, and you have to add cells linearly to add either capacity or duration. There is no architectural advantage to building longer-duration lithium systems.

Worse, the revenue model breaks down. A four-hour battery cycled daily can capture two profitable arbitrage spreads per day — charging during the midday solar surplus and discharging during the evening peak. A twelve-hour battery cycled daily captures only one profitable spread per day, and the marginal value of the additional duration is small. To pay back its higher capital cost, a long-duration lithium battery would need to capture much larger spreads than the market typically offers. In most markets, those spreads do not exist often enough to justify the investment.

This is why, despite the extraordinary growth of grid-scale lithium storage, the average installed duration in the United States has remained stuck around four hours since 2020. The technology is excellent at what it does. It is not the right technology for the job that comes next.

Several technologies are competing to fill the long-duration gap. None has yet won. Each has different strengths, different weaknesses, and different cost trajectories.

Iron-air batteries use a cell chemistry based on the rusting of iron, an element so abundant that material cost is essentially zero. Form Energy, the leading developer, has projected installed costs of roughly $20 per kilowatt-hour for 100-hour systems; a figure roughly an order of magnitude below lithium. The first commercial deployment, a 1-megawatt, 100-hour system in Minnesota, came online in 2024. Whether the technology can scale at the projected cost is the central open question. If it can, iron-air becomes the dominant long-duration technology and natural gas becomes optional rather than necessary.

Vanadium redox flow batteries decouple power and energy by storing energy in liquid electrolyte tanks separate from the power-conversion stack. This means duration can be extended cheaply by adding more electrolyte. Installed costs for 12-hour vanadium systems are currently around $400 per kilowatt-hour and falling. Vanadium has been deployed at hundred-megawatt scale in China and is gaining traction in Australia. The constraint is vanadium supply, which is concentrated in China, Russia, and South Africa; not unlike the cobalt and nickel constraints that lithium faces.

Sodium-ion batteries offer chemistry similar to lithium-ion but with cheaper, more abundant materials. CATL, BYD, and several Western developers have begun commercial deployment. Energy density is lower, which makes sodium-ion a poor fit for electric vehicles but a reasonable fit for stationary storage. Installed costs are projected to reach lithium-iron-phosphate parity by 2027 or 2028, with longer-duration variants potentially undercutting lithium for eight- to twelve-hour applications.

Thermal storage; heating crushed rock, molten salt, or other media with renewable electricity and discharging via a heat engine, offers another path. Antora Energy, Rondo, and Heliogen are among the leading developers. The technology is mechanically simple, the materials are abundant, and the round-trip efficiency is lower than batteries (roughly 40 to 50 percent) but the capital cost per kilowatt-hour can be very low. The economics work for industrial heat applications today; the economics for power-only applications are still being proven.

Gravity-based and mechanical systems; pumped hydro, compressed air, gravity batteries; round out the field. Pumped hydro remains the largest installed long-duration storage technology globally and will continue to be deployed where geography permits. The newer mechanical systems are promising but have not yet demonstrated cost competitiveness at scale.

The contest among these technologies will not be settled by which has the best laboratory performance. It will be settled by which can deploy at scale, hit cost reduction curves, and prove durability over multi-decade operating lives. By 2030, we will have a clearer picture of which technologies have crossed those thresholds and which have not. The technologies that succeed will reshape the grid in ways that are hard to overstate. The technologies that fail will be remembered as cautionary tales of promising chemistries that could not survive the journey from pilot to commercial deployment.

What is already clear is the prize. A grid that has economically viable long-duration storage is a grid that does not need natural gas as a seasonal backstop. It is a grid that can be powered, reliably and affordably, almost entirely by renewable generation. It is the grid that the climate requires and that the economics of the technology, if the cost curves bend the right way, can deliver. The next five years will tell us whether that grid is going to exist.