Why batteries are changing the electricity market

Electricity is unusual among the things you buy: for the most part it has to be generated at the same moment it is used.

That is not quite an absolute rule, and it never was. The United States has been storing electricity at scale since long before anyone put a lithium battery on a grid. What has changed is how, and how much.

Storage is old. This kind of storage is new.

The form of grid storage the US has leaned on longest is pumped storage hydropower. Writing in 2019, the EIA called pumped storage "the largest source of electricity storage technology used in the United States, both in terms of capacity and number of plants." Battery capacity has since passed it.

The EIA describes how it works simply: water is pumped into an uphill reservoir "during times of relatively low electricity demand and low electricity prices, such as during the night," and when demand is high the water "flows downhill from the reservoir through hydroelectric generators at a dam."

It is a battery made of water and elevation, and it is not new. Pumped storage plants in the US "were built primarily between 1960 and 1990," and nearly half of the capacity still running was built in the 1970s.

Pumped storage needs two reservoirs at different elevations, which is a demanding thing to ask of a site. Batteries do not need the elevation. That is the change — though how much of pumped storage's long plateau is explained by geography, as against cost, permitting or anything else, is not something these sources settle.

How much battery capacity there is now

The EIA tracks every utility-scale generator in the country through a monthly survey of operators. Its figures for battery storage:

• By the end of 2025, the US power system had 43.6 gigawatts of operational battery storage capacity.
• In the first six months of 2026, operators added another 8.3 gigawatts, reaching nearly 52 gigawatts of nameplate capacity.
• Over the three years to mid-2026, utility-scale battery capacity grew at an annual average rate of 70%.
• Operators have told the EIA they expect to bring an additional 54 gigawatts online over the following two and a half years.

For the same year, 2025, the EIA reports that the United States had "about 23,156 MW of total pumped-storage hydroelectricity generation capacity in 18 states" — roughly 23.2 gigawatts.

So at the end of 2025 the battery fleet was rated at nearly twice the capacity of the pumped-storage fleet that took three decades to build.

Two things to be careful about before reading too much into that

The two numbers are measured on different bases. The battery figure is nameplate capacity, which EIA defines as "the maximum rated output of a generator, prime mover, or other electric power production equipment under specific conditions designated by the manufacturer." The pumped-storage figure is reported as net summer capacity: "the maximum output, commonly expressed in megawatts (MW), that generating equipment can supply to system load, as demonstrated by a multi-hour test, at the time of summer peak demand."

One is a rating from the manufacturer. The other is a demonstrated figure under summer conditions. They are close cousins rather than the same measurement, so treat the comparison as an indication of scale, not a precise ratio.

Gigawatts measure power, not energy. This is the distinction that trips people up most often with storage.

A gigawatt is a rate — how much electricity something can deliver at a given moment. It says nothing about how long it can keep delivering it. That depends on how much energy is stored, which is measured in gigawatt-hours.

Two storage fleets rated at the same number of gigawatts can behave completely differently: one might sustain that output for an hour, the other for ten. None of the capacity figures above tells you the discharge duration of either fleet, and neither does this article.

What operators actually use batteries for

A battery on a grid can do several different jobs, and which job it is doing determines what it does to prices. The EIA asks operators what each battery is used for, and the answers have shifted.

In 2021, the leading application was frequency regulation — the constant small adjustments that hold the grid at 60 cycles per second. Batteries suit that work because they can change their output quickly. Frequency regulation was among the uses of 63% of battery capacity; arbitrage was among the uses of 58%.

By the EIA's September 2025 analysis, that order had reversed. Arbitrage was among the uses of 66% of all utility-scale battery capacity, and was the primary use of 41% of it. Frequency regulation was the primary use of 24%.

Arbitrage, in the EIA's own words, "involves buying electricity when prices are relatively low and selling that electricity when prices are high."

A battery doing arbitrage is not generating anything. It is moving energy from one hour to another.

What moving energy between hours does to prices

Grid operators meet demand by calling on generators roughly in order of cost, cheapest first, until supply matches demand. When demand is low, only the cheap end of that stack is needed. When demand is high, the operator has to reach further up it.

Storage acts on both ends of that pattern. Charging adds demand during the cheaper hours. Discharging adds supply during the more expensive ones, so the generators that would otherwise have been called on face competition.

It does not follow that storage puts a ceiling on peak prices, and this article is not claiming that it does. Several things cut against it:

• A battery can only discharge for as long as its stored energy lasts. A peak longer than that duration is a peak it cannot cover to the end.
• How much difference storage makes depends on how much of it there is relative to the size of the peak, and that ratio varies enormously between regions.
• Market rules, transmission limits and local conditions all affect whether stored energy can reach the demand that wants it.
• Prices can still rise sharply when supply is genuinely scarce. The EIA notes that extreme temperatures raise demand and that the resulting increases "can increase fuel and electricity prices."

The honest version is narrower and still worth saying: storage introduces competition into hours that previously had less of it. What that competition is worth, in any particular market, is a question about that market.

Storage is not free energy

Every storage system returns less electricity than it takes in.

The EIA measures this as round-trip efficiency, "the percentage of electricity put into storage that is later retrieved."

Its published figures are fleet-wide averages for 2019, not current numbers and not the rating of any individual installation: in that year the US utility-scale battery fleet averaged a monthly round-trip efficiency of 82%, and pumped-storage facilities averaged 79%.

For pumped storage specifically, the EIA notes that most systems "require 15% to 30% more electricity to pump water uphill than what the water generates when it flows back downhill."

This is also why the EIA reports both gross and net generation for storage, with net figures often negative.

The practical consequence is a real economic test: storing energy only pays when the price gap between the cheap hours and the expensive hours is wider than the losses plus the cost of the equipment.

What is confirmed, and what is not

Confirmed, from federal data: every figure above. The battery capacity numbers and growth rate, the 2025 pumped-storage capacity and the period the pumped-storage fleet was built, both capacity definitions, the application shares for both 2021 and 2025, and both 2019 round-trip efficiency averages are all published by the EIA and linked below.

Analysis, not fact: the description of how calling on generators in cost order translates into an hourly price pattern is the standard account of how these markets work. It is a simplification, and real markets add constraints that can change which generator sets the price. The reading that more storage means more competition in expensive hours follows from that account; it is reasoning, not a measurement, and the limits on it are listed above. The suggestion that site requirements held pumped storage back is an inference from what pumped storage needs, not a finding in any source cited here. The battery-to-pumped-storage comparison is between two figures EIA reports on different bases, which is stated where the comparison is made.

Not known from these sources: what any of this does to a specific household's bill. Retail rates are set by state regulators and utilities, and the relationship between wholesale prices and a residential tariff varies by jurisdiction. Nothing above is a forecast of anybody's bill.